Merge remote-tracking branch 'upstream/master'

6 files changed, 3438 insertions, 3438 deletions

diff --git a/mp/src/utils/phonemeextractor/extractor_utils.cpp b/mp/src/utils/phonemeextractor/extractor_utils.cpp
index ba927f04..8b8ff98b 100644
--- a/mp/src/utils/phonemeextractor/extractor_utils.cpp
+++ b/mp/src/utils/phonemeextractor/extractor_utils.cpp
@@ -1,28 +1,28 @@
-//========= Copyright Valve Corporation, All rights reserved. ============//

-//

-// Purpose: 

-//

-//=============================================================================//

-#include <windows.h>

-#include <stdio.h>

-

-//-----------------------------------------------------------------------------

-// Purpose: converts an english string to unicode

-//-----------------------------------------------------------------------------

-int ConvertANSIToUnicode(const char *ansi, wchar_t *unicode, int unicodeBufferSize)

-{

-	return ::MultiByteToWideChar(CP_ACP, 0, ansi, -1, unicode, unicodeBufferSize);

-}

-

-char *va( const char *fmt, ... )

-{

-	va_list args;

-	static char output[4][1024];

-	static int outbuffer = 0;

-

-	outbuffer++;

-	va_start( args, fmt );

-	vprintf( fmt, args );

-	vsprintf( output[ outbuffer & 3 ], fmt, args );

-	return output[ outbuffer & 3 ];

+//========= Copyright Valve Corporation, All rights reserved. ============//
+//
+// Purpose: 
+//
+//=============================================================================//
+#include <windows.h>
+#include <stdio.h>
+
+//-----------------------------------------------------------------------------
+// Purpose: converts an english string to unicode
+//-----------------------------------------------------------------------------
+int ConvertANSIToUnicode(const char *ansi, wchar_t *unicode, int unicodeBufferSize)
+{
+	return ::MultiByteToWideChar(CP_ACP, 0, ansi, -1, unicode, unicodeBufferSize);
+}
+
+char *va( const char *fmt, ... )
+{
+	va_list args;
+	static char output[4][1024];
+	static int outbuffer = 0;
+
+	outbuffer++;
+	va_start( args, fmt );
+	vprintf( fmt, args );
+	vsprintf( output[ outbuffer & 3 ], fmt, args );
+	return output[ outbuffer & 3 ];
 }
\ No newline at end of file
diff --git a/mp/src/utils/phonemeextractor/phonemeextractor.cpp b/mp/src/utils/phonemeextractor/phonemeextractor.cpp
index 8dfc8439..271f1850 100644
--- a/mp/src/utils/phonemeextractor/phonemeextractor.cpp
+++ b/mp/src/utils/phonemeextractor/phonemeextractor.cpp
@@ -1,1425 +1,1425 @@
-//========= Copyright Valve Corporation, All rights reserved. ============//

-//

-// Purpose: 

-//

-// $NoKeywords: $

-//

-//=============================================================================//

-// extracephonemes.cpp : Defines the entry point for the console application.

-//

-#define PROTECTED_THINGS_DISABLE

-

-#include "tier0/wchartypes.h"

-#include <stdio.h>

-#include <windows.h>

-#include <tchar.h>

-#include "sphelper.h"

-#include "spddkhlp.h"

-// ATL Header Files

-#include <atlbase.h>

-// Face poser and util includes

-#include "utlvector.h"

-#include "phonemeextractor/PhonemeExtractor.h"

-#include "PhonemeConverter.h"

-#include "sentence.h"

-#include "tier0/dbg.h"

-#include "tier0/icommandline.h"

-#include "filesystem.h"

-

-// Extract phoneme grammar id

-#define EP_GRAM_ID			101

-// First rule of dynamic sentence rule set

-#define DYN_SENTENCERULE	102

-// # of milliseconds to allow for processing before timeout

-#define SR_WAVTIMEOUT		4000

-// Weight tag for rule to rule word/rule transitions

-#define CONFIDENCE_WEIGHT	0.0f

-

-//#define LOGGING		1

-#define LOGFILE		"c:\\fp.log"

-

-void LogReset( void )

-{

-#if LOGGING

-	FILE *fp = fopen( LOGFILE, "w" );

-	if ( fp )

-		fclose( fp );

-#endif

-}

-

-char *va( const char *fmt, ... );

-

-//-----------------------------------------------------------------------------

-// Purpose: 

-// Input  : *words - 

-//-----------------------------------------------------------------------------

-void LogWords( CSentence& sentence )

-{

-	Log( "Wordcount == %i\n", sentence.m_Words.Size() );

-

-	for ( int i = 0; i < sentence.m_Words.Size(); i++ )

-	{

-		const CWordTag *w = sentence.m_Words[ i ];

-		Log( "Word %s %u to %u\n", w->GetWord(), w->m_uiStartByte, w->m_uiEndByte );

-	}

-}

-

-//-----------------------------------------------------------------------------

-// Purpose: 

-// Input  : *phonemes - 

-//-----------------------------------------------------------------------------

-void LogPhonemes( CSentence& sentence )

-{

-	return;

-

-	Log( "Phonemecount == %i\n", sentence.CountPhonemes() );

-

-	for ( int i = 0; i < sentence.m_Words.Size(); i++ )

-	{

-		const CWordTag *w = sentence.m_Words[ i ];

-

-		for ( int j = 0; j < w->m_Phonemes.Size(); j++ )

-		{

-			const CPhonemeTag *p = w->m_Phonemes[ j ];

-			Log( "Phoneme %s %u to %u\n", p->GetTag(), p->m_uiStartByte, p->m_uiEndByte );

-		}

-	}

-}

-

-#define NANO_CONVERT 10000000.0f;

-

-//-----------------------------------------------------------------------------

-// Purpose: Walk list of words and phonemes and create phoneme tags in CSentence object

-//  FIXME:  Right now, phonemes are assumed to evenly space out across a word.

-// Input  : *converter - 

-//			result - 

-//			sentence - 

-//-----------------------------------------------------------------------------

-void EnumeratePhonemes( ISpPhoneConverter *converter, const ISpRecoResult* result, CSentence& sentence )

-{

-	USES_CONVERSION;

-

-	// Grab access to element container

-	ISpPhrase *phrase = ( ISpPhrase * )result;

-	if ( !phrase )

-		return;

-

-    SPPHRASE *pElements;

-	if ( !SUCCEEDED( phrase->GetPhrase( &pElements ) ) )

-		return;

-

-	// Only use it if it's better/same size as what we already had on-hand

-	if ( pElements->Rule.ulCountOfElements > 0 )

-		//(unsigned int)( sentence.m_Words.Size() - sentence.GetWordBase() ) )

-	{

-		sentence.ResetToBase();

-

-		// Walk list of words

-		for ( ULONG i = 0; i < pElements->Rule.ulCountOfElements; i++ )

-		{

-			unsigned int wordstart, wordend;

-

-			// Get start/end sample index

-			wordstart	= pElements->pElements[i].ulAudioStreamOffset + (unsigned int)pElements->ullAudioStreamPosition;

-			wordend		= wordstart + pElements->pElements[i].ulAudioSizeBytes;

-

-			// Create word tag

-			CWordTag *w = new CWordTag( W2T( pElements->pElements[i].pszDisplayText ) );

-			Assert( w );

-			w->m_uiStartByte = wordstart;

-			w->m_uiEndByte   = wordend;

-

-			sentence.AddWordTag( w );

-

-			// Count # of phonemes in this word

-			SPPHONEID pstr[ 2 ];

-			pstr[ 1 ] = 0;

-			WCHAR wszPhoneme[ SP_MAX_PRON_LENGTH ];

-

-			const SPPHONEID *current;

-			SPPHONEID phoneme;

-			current = pElements->pElements[i].pszPronunciation;

-			float total_weight = 0.0f;

-			while ( 1 )

-			{

-				phoneme = *current++;

-				if ( !phoneme )

-					break;

-

-				pstr[ 0 ] = phoneme;

-				wszPhoneme[ 0 ] = L'\0';

-

-				converter->IdToPhone( pstr, wszPhoneme );

-

-				total_weight += WeightForPhoneme( W2A( wszPhoneme ) );

-			}

-

-			current = pElements->pElements[i].pszPronunciation;

-

-			// Decide # of bytes/phoneme weight

-			float psize = 0;

-			if ( total_weight )

-			{

-				psize = ( wordend - wordstart ) / total_weight;

-			}

-

-			int number = 0;

-

-			// Re-walk the phoneme list and create true phoneme tags

-			float startWeight = 0.0f;

-			while ( 1 )

-			{

-				phoneme = *current++;

-				if ( !phoneme )

-					break;

-

-				pstr[ 0 ] = phoneme;

-				wszPhoneme[ 0 ] = L'\0';

-

-				converter->IdToPhone( pstr, wszPhoneme );

- 

-				CPhonemeTag *p = new CPhonemeTag( W2A( wszPhoneme ) );

-				Assert( p );

-				

-				float weight = WeightForPhoneme( W2A( wszPhoneme ) );

-

-				p->m_uiStartByte = wordstart + (int)( startWeight * psize );

-				p->m_uiEndByte	 = p->m_uiStartByte + (int)( psize * weight );

-

-				startWeight += weight;

-

-				// Convert to IPA phoneme code

-				p->SetPhonemeCode( TextToPhoneme( p->GetTag() ) );

-

-				sentence.AddPhonemeTag( w, p );

-

-				number++;

-			}

-		}	

-	}

-

-	// Free memory

-    ::CoTaskMemFree(pElements);

-}

-

-//-----------------------------------------------------------------------------

-// Purpose: Create rules for each word in the reference sentence

-//-----------------------------------------------------------------------------

-typedef struct

-{

-	int					ruleId;

-	SPSTATEHANDLE		hRule;

-	CSpDynamicString	word;

-	char				plaintext[ 256 ];

-} WORDRULETYPE;

-

-//-----------------------------------------------------------------------------

-// Purpose: Creates start for word of sentence

-// Input  : cpRecoGrammar - 

-//			*root - 

-//			*rules - 

-//			word - 

-//-----------------------------------------------------------------------------

-void AddWordRule( ISpRecoGrammar* cpRecoGrammar, SPSTATEHANDLE *root, CUtlVector< WORDRULETYPE > *rules, CSpDynamicString& word )

-{

-	USES_CONVERSION;

-	HRESULT hr;

-	WORDRULETYPE *newrule;

-

-	int idx = (*rules).AddToTail();

-

-	newrule = &(*rules)[ idx ];

-

-	newrule->ruleId = DYN_SENTENCERULE + idx + 1;

-	newrule->word = word;

-

-	strcpy( newrule->plaintext, W2T( word ) );

-

-	// Create empty rule

-	hr = cpRecoGrammar->CreateNewState( *root, &newrule->hRule );

-	Assert( !FAILED( hr ) );

-}

-

-//-----------------------------------------------------------------------------

-// Purpose: 

-// Input  : cpRecoGrammar - 

-//			*from - 

-//			*to - 

-//-----------------------------------------------------------------------------

-void AddWordTransitionRule( ISpRecoGrammar* cpRecoGrammar, WORDRULETYPE *from, WORDRULETYPE *to )

-{

-	USES_CONVERSION;

-

-	HRESULT hr;

-	Assert( from );

-

-	if ( from && !to )

-	{

-		OutputDebugString( va( "Transition from %s to TERM\r\n", from->plaintext ) );

-	}

-	else

-	{

-		OutputDebugString( va( "Transition from %s to %s\r\n", from->plaintext, to->plaintext ) );

-	}

-

-	hr = cpRecoGrammar->AddWordTransition( from->hRule, to ? to->hRule : NULL, (WCHAR *)from->word, NULL, SPWT_LEXICAL, CONFIDENCE_WEIGHT, NULL );

-	Assert( !FAILED( hr ) );

-}

-

-//-----------------------------------------------------------------------------

-// Purpose: 

-// Input  : cpRecoGrammar - 

-//			*from - 

-//			*to - 

-//-----------------------------------------------------------------------------

-void AddOptionalTransitionRule( ISpRecoGrammar* cpRecoGrammar, WORDRULETYPE *from, WORDRULETYPE *to )

-{

-	USES_CONVERSION;

-

-	HRESULT hr;

-	Assert( from );

-

-	if ( from && !to )

-	{

-		OutputDebugString( va( "Opt transition from %s to TERM\r\n", from->plaintext ) );

-	}

-	else

-	{

-		OutputDebugString( va( "Opt transition from %s to %s\r\n", from->plaintext, to->plaintext ) );

-	}

-

-	hr = cpRecoGrammar->AddWordTransition( from->hRule, to ? to->hRule : NULL, NULL, NULL, SPWT_LEXICAL, CONFIDENCE_WEIGHT, NULL );

-	Assert( !FAILED( hr ) );

-}

-

-#define MAX_WORD_SKIP 1

-//-----------------------------------------------------------------------------

-// Purpose: Links together all word rule states into a sentence rule CFG

-// Input  : singleword - 

-//			cpRecoGrammar - 

-//			*root - 

-//			*rules - 

-//-----------------------------------------------------------------------------

-bool BuildRules( ISpRecoGrammar* cpRecoGrammar, SPSTATEHANDLE *root, CUtlVector< WORDRULETYPE > *rules )

-{

-	HRESULT hr;

-	WORDRULETYPE *rule, *next;

-

-	int numrules = (*rules).Size();

-

-	rule = &(*rules)[ 0 ];

-

-	// Add transition

-	hr = cpRecoGrammar->AddWordTransition( *root, rule->hRule, NULL, NULL, SPWT_LEXICAL, CONFIDENCE_WEIGHT, NULL );

-	Assert( !FAILED( hr ) );

-

-	for ( int i = 0; i < numrules; i++ )

-	{

-		rule = &(*rules)[ i ];

-		if ( i < numrules - 1 )

-		{

-			next = &(*rules)[ i + 1 ];

-		}

-		else

-		{

-			next = NULL;

-		}

-

-		AddWordTransitionRule( cpRecoGrammar, rule, next );

-	}

-

-	if ( numrules > 1 )

-	{

-		for ( int skip = 1; skip <= min( MAX_WORD_SKIP, numrules ); skip++ )

-		{

-			OutputDebugString( va( "Opt transition from Root to %s\r\n", (*rules)[ 0 ].plaintext ) );

-

-			hr = cpRecoGrammar->AddWordTransition( *root, (*rules)[ 0 ].hRule, NULL, NULL, SPWT_LEXICAL, CONFIDENCE_WEIGHT, NULL );

-

-			// Now build rules where you can skip 1 to N intervening words

-			for ( int i = 1; i < numrules; i++ )

-			{

-				// Start at the beginning?

-				rule = &(*rules)[ i ];	

-				if ( i < numrules - skip )

-				{

-					next = &(*rules)[ i + skip ];

-				}

-				else

-				{

-					continue;

-				}

-

-				// Add transition

-				AddOptionalTransitionRule( cpRecoGrammar, rule, next );

-			}

-

-			// Go from final rule to end point

-			AddOptionalTransitionRule( cpRecoGrammar, rule, NULL );

-		}

-	}

-

-	// Store it

-	hr = cpRecoGrammar->Commit(NULL);

-	if ( FAILED( hr ) )

-		return false;

-

-	return true;

-}

-

-//-----------------------------------------------------------------------------

-// Purpose: Debugging, prints alternate list if one is created

-// Input  : cpResult - 

-//			(*pfnPrint - 

-//-----------------------------------------------------------------------------

-void PrintAlternates( ISpRecoResult* cpResult, void (*pfnPrint)( const char *fmt, ... ) )

-{

-	ISpPhraseAlt *rgPhraseAlt[ 32 ];

-	memset( rgPhraseAlt, 0, sizeof( rgPhraseAlt ) );

-

-	ULONG ulCount;

-	

-	ISpPhrase *phrase = ( ISpPhrase * )cpResult;

-	if ( phrase )

-	{

-		SPPHRASE *pElements;

-		if ( SUCCEEDED( phrase->GetPhrase( &pElements ) ) )

-		{

-			if ( pElements->Rule.ulCountOfElements > 0 )

-			{

-				HRESULT hr = cpResult->GetAlternates(

-					pElements->Rule.ulFirstElement,

-					pElements->Rule.ulCountOfElements, 

-					32,

-					rgPhraseAlt,

-					&ulCount);

-				

-				Assert( !FAILED( hr ) );

-				

-				for ( ULONG r = 0 ; r < ulCount; r++ )

-				{

-					CSpDynamicString dstrText;

-					hr = rgPhraseAlt[ r ]->GetText( (ULONG)SP_GETWHOLEPHRASE, (ULONG)SP_GETWHOLEPHRASE, TRUE, &dstrText, NULL);

-					Assert( !FAILED( hr ) );

-

-					pfnPrint( "[ ALT ]" );

-					pfnPrint( dstrText.CopyToChar() );

-					pfnPrint( "\r\n" );

-				}

-			}

-		}

-		

-	}

-

-	for ( int i = 0; i < 32; i++ )

-	{

-		if ( rgPhraseAlt[ i ] )

-		{

-			rgPhraseAlt[ i ]->Release();

-			rgPhraseAlt[ i ] = NULL;

-		}

-	}

-}

-

-void PrintWordsAndPhonemes( CSentence& sentence, void (*pfnPrint)( const char *fmt, ... ) )

-{

-	char sz[ 256 ];

-	int i;

-

-	pfnPrint( "WORDS\r\n\r\n" );

-

-	for ( i = 0 ; i < sentence.m_Words.Size(); i++ )

-	{

-		CWordTag *word = sentence.m_Words[ i ];

-		if ( !word )

-			continue;

-

-		sprintf( sz, "<%u - %u> %s\r\n", 

-			word->m_uiStartByte, word->m_uiEndByte, word->GetWord() );

-

-		pfnPrint( sz );

-

-		for ( int j = 0 ; j < word->m_Phonemes.Size(); j++ )

-		{

-			CPhonemeTag *phoneme = word->m_Phonemes[ j ];

-			if ( !phoneme )

-				continue;

-

-			sprintf( sz, "  <%u - %u> %s\r\n", 

-				phoneme->m_uiStartByte, phoneme->m_uiEndByte, phoneme->GetTag() );

-

-			pfnPrint( sz );

-		}

-	}

-

-	pfnPrint( "\r\n" );

-}

-

-//-----------------------------------------------------------------------------

-// Purpose: Given a wave file and a string of words "text", creates a CFG from the

-//  sentence and stores the resulting words/phonemes in CSentence

-// Input  : *wavname - 

-//			text - 

-//			sentence - 

-//			(*pfnPrint - 

-// Output : SR_RESULT

-//-----------------------------------------------------------------------------

-SR_RESULT ExtractPhonemes( const char *wavname, CSpDynamicString& text, CSentence& sentence, void (*pfnPrint)( const char *fmt, ...) )

-{

-	// Assume failure

-	SR_RESULT result = SR_RESULT_ERROR;

-

-	if ( text.Length() <= 0 )

-	{

-		pfnPrint( "Error:  no rule / text specified\n" );

-		return result;

-	}

-

-	USES_CONVERSION;

-	HRESULT hr;

-	

-	CUtlVector < WORDRULETYPE > wordRules;

-

-	CComPtr<ISpStream> cpInputStream;

-	CComPtr<ISpRecognizer> cpRecognizer;

-	CComPtr<ISpRecoContext> cpRecoContext;

-	CComPtr<ISpRecoGrammar> cpRecoGrammar;

-	CComPtr<ISpPhoneConverter>  cpPhoneConv;

-    

-	// Create basic SAPI stream object

-	// NOTE: The helper SpBindToFile can be used to perform the following operations

-	hr = cpInputStream.CoCreateInstance(CLSID_SpStream);

-	if ( FAILED( hr ) )

-	{

-		pfnPrint( "Error:  SAPI 5.1 Stream object not installed?\n" );

-		return result;

-	}

-

-	CSpStreamFormat sInputFormat;

-	

-	// setup stream object with wav file MY_WAVE_AUDIO_FILENAME

-	//   for read-only access, since it will only be access by the SR engine

-	hr = cpInputStream->BindToFile(

-		T2W(wavname),

-		SPFM_OPEN_READONLY,

-		NULL,

-		sInputFormat.WaveFormatExPtr(),

-		SPFEI_ALL_EVENTS );

-

-	if ( FAILED( hr ) )

-	{

-		pfnPrint( "Error: couldn't open wav file %s\n", wavname );

-		return result;

-	}

-	

-	// Create in-process speech recognition engine

-	hr = cpRecognizer.CoCreateInstance(CLSID_SpInprocRecognizer);

-	if ( FAILED( hr ) )

-	{

-		pfnPrint( "Error:  SAPI 5.1 In process recognizer object not installed?\n" );

-		return result;

-	}

-

-	// Create recognition context to receive events

-	hr = cpRecognizer->CreateRecoContext(&cpRecoContext);

-	if ( FAILED( hr ) )

-	{

-		pfnPrint( "Error:  SAPI 5.1 Unable to create recognizer context\n" );

-		return result;

-	}

-	

-	// Create a grammar

-	hr = cpRecoContext->CreateGrammar( EP_GRAM_ID, &cpRecoGrammar );

-	if ( FAILED( hr ) )

-	{

-		pfnPrint( "Error:  SAPI 5.1 Unable to create recognizer grammar\n" );

-		return result;

-	}

-

-	LANGID englishID = 0x409; // 1033 decimal

-

-	bool userSpecified = false;

-	LANGID langID = SpGetUserDefaultUILanguage();

-

-	// Allow commandline override

-	if ( CommandLine()->FindParm( "-languageid" ) != 0 )

-	{

-		userSpecified = true;

-		langID = CommandLine()->ParmValue( "-languageid", langID );

-	}

-

-	// Create a phoneme converter ( so we can convert to IPA codes )

-	hr = SpCreatePhoneConverter( langID, NULL, NULL, &cpPhoneConv );

-	if ( FAILED( hr ) )

-	{

-		if ( langID != englishID )

-		{

-			if ( userSpecified )

-			{

-				pfnPrint( "Warning:  SAPI 5.1 Unable to create phoneme converter for command line override -languageid %i\n", langID );

-			}

-			else

-			{

-				pfnPrint( "Warning:  SAPI 5.1 Unable to create phoneme converter for default UI language %i\n",langID );

-			}

-

-			// Try english!!!

-			langID = englishID;

-			hr = SpCreatePhoneConverter( langID, NULL, NULL, &cpPhoneConv );

-		}

-

-		if ( FAILED( hr ) )

-		{

-			pfnPrint( "Error:  SAPI 5.1 Unable to create phoneme converter for English language id %i\n", langID );

-			return result;

-		}

-		else

-		{

-			pfnPrint( "Note:  SAPI 5.1 Falling back to use english -languageid %i\n", langID );

-		}

-	}

-	else if ( userSpecified )

-	{

-		pfnPrint( "Note:  SAPI 5.1 Using user specified -languageid %i\n",langID );

-	}

-

-	SPSTATEHANDLE hStateRoot;

-	// create/re-create Root level rule of grammar

-	hr = cpRecoGrammar->GetRule(L"Root", 0, SPRAF_TopLevel | SPRAF_Active, TRUE, &hStateRoot);

-	if ( FAILED( hr ) )

-	{

-		pfnPrint( "Error:  SAPI 5.1 Unable to create root rule\n" );

-		return result;

-	}

-

-	// Inactivate it so we can alter it

-	hr = cpRecoGrammar->SetRuleState( NULL, NULL, SPRS_INACTIVE );

-	if ( FAILED( hr ) )

-	{

-		pfnPrint( "Error:  SAPI 5.1 Unable to deactivate grammar rules\n" );

-		return result;

-	}

-

-	// Create the rule set from the words in text

-	{

-		CSpDynamicString currentWord;

-		WCHAR *pos = ( WCHAR * )text;

-		WCHAR str[ 2 ];

-		str[1]= 0;

-

-		while ( *pos )

-		{

-			if ( *pos == L' ' /*|| *pos == L'.' || *pos == L'-'*/ )

-			{

-				// Add word to rule set

-				if ( currentWord.Length() > 0 )

-				{

-					AddWordRule( cpRecoGrammar, &hStateRoot, &wordRules, currentWord );

-					currentWord.Clear();

-				}

-				pos++;

-				continue;

-			}

-

-			// Skip anything that's inside a [ xxx ] pair.

-			if ( *pos == L'[' )

-			{

-				while ( *pos && *pos != L']' )

-				{

-					pos++;

-				}

-

-				if ( *pos )

-				{

-					pos++;

-				}

-				continue;

-			}

-

-			str[ 0 ] = *pos;

-

-			currentWord.Append( str );

-			pos++;

-		}

-

-		if ( currentWord.Length() > 0 )

-		{

-			AddWordRule( cpRecoGrammar, &hStateRoot, &wordRules, currentWord );

-		}

-

-		if ( wordRules.Size() <= 0 )

-		{

-			pfnPrint( "Error:  Text %s contained no usable words\n", text );

-			return result;

-		}

-

-		// Build all word to word transitions in the grammar

-		if ( !BuildRules( cpRecoGrammar, &hStateRoot, &wordRules ) )

-		{

-			pfnPrint( "Error:  Rule set for %s could not be generated\n", text );

-			return result;

-		}

-	}

-

-	// check for recognitions and end of stream event

-	const ULONGLONG ullInterest = 

-		SPFEI(SPEI_RECOGNITION) | SPFEI(SPEI_END_SR_STREAM) | SPFEI(SPEI_FALSE_RECOGNITION) | 

-		SPFEI(SPEI_PHRASE_START ) | SPFEI(SPEI_HYPOTHESIS ) | SPFEI(SPEI_INTERFERENCE) ;

-	hr = cpRecoContext->SetInterest( ullInterest, ullInterest );

-	if ( FAILED( hr ) )

-	{

-		pfnPrint( "Error:  SAPI 5.1 Unable to set interest level\n" );

-		return result;

-	}	

-	// use Win32 events for command-line style application

-	hr = cpRecoContext->SetNotifyWin32Event();

-	if ( FAILED( hr ) )

-	{

-		pfnPrint( "Error:  SAPI 5.1 Unable to set win32 notify event\n" );

-		return result;

-	}

-	// connect wav input to recognizer

-	// SAPI will negotiate mismatched engine/input audio formats using system audio codecs, so second parameter is not important - use default of TRUE

-	hr = cpRecognizer->SetInput(cpInputStream, TRUE);

-	if ( FAILED( hr ) )

-	{

-		pfnPrint( "Error:  SAPI 5.1 Unable to associate input stream\n" );

-		return result;

-	}	

-

-	// Activate the CFG ( rather than using dictation )

-	hr = cpRecoGrammar->SetRuleState( NULL, NULL, SPRS_ACTIVE );

-	if ( FAILED( hr ) )

-	{

-		switch ( hr )

-		{

-		case E_INVALIDARG:

-			pfnPrint( "pszName is invalid or bad. Alternatively, pReserved is non-NULL\n" );

-			break;

-		case SP_STREAM_UNINITIALIZED:

-			pfnPrint( "ISpRecognizer::SetInput has not been called with the InProc recognizer\n" );

-			break;

-		case SPERR_UNINITIALIZED:

-			pfnPrint( "The object has not been properly initialized.\n");

-			break;

-		case SPERR_UNSUPPORTED_FORMAT:

-			pfnPrint( "Audio format is bad or is not recognized. Alternatively, the device driver may be busy by another application and cannot be accessed.\n" );

-			break;

-		case SPERR_NOT_TOPLEVEL_RULE:

-			pfnPrint( "The rule pszName exists, but is not a top-level rule.\n" );

-			break;

-		default:

-			pfnPrint( "Unknown error\n" );

-			break;

-		}

-		pfnPrint( "Error:  SAPI 5.1 Unable to activate rule set\n" );

-		return result;

-	}

-

-	// while events occur, continue processing

-	// timeout should be greater than the audio stream length, or a reasonable amount of time expected to pass before no more recognitions are expected in an audio stream

-	BOOL fEndStreamReached = FALSE;

-	while (!fEndStreamReached && S_OK == cpRecoContext->WaitForNotifyEvent( SR_WAVTIMEOUT ))

-	{

-		CSpEvent spEvent;

-		// pull all queued events from the reco context's event queue

-		

-		while (!fEndStreamReached && S_OK == spEvent.GetFrom(cpRecoContext))

-		{

-			// Check event type

-			switch (spEvent.eEventId)

-			{

-			case SPEI_INTERFERENCE:

-				{

-					SPINTERFERENCE interference = spEvent.Interference();

-

-					switch ( interference )

-					{

-					case SPINTERFERENCE_NONE:

-						pfnPrint( "[ I None ]\r\n" );

-						break;

-					case SPINTERFERENCE_NOISE:

-						pfnPrint( "[ I Noise ]\r\n" );

-						break;

-					case SPINTERFERENCE_NOSIGNAL:

-						pfnPrint( "[ I No Signal ]\r\n" );

-						break;

-					case SPINTERFERENCE_TOOLOUD:

-						pfnPrint( "[ I Too Loud ]\r\n" );

-						break;

-					case SPINTERFERENCE_TOOQUIET:

-						pfnPrint( "[ I Too Quiet ]\r\n" );

-						break;

-					case SPINTERFERENCE_TOOFAST:

-						pfnPrint( "[ I Too Fast ]\r\n" );

-						break;

-					case SPINTERFERENCE_TOOSLOW:

-						pfnPrint( "[ I Too Slow ]\r\n" );

-						break;

-					default:

-						break;

-					}

-				}

-				break;

-			case SPEI_PHRASE_START:

-				pfnPrint( "Phrase Start\r\n" );

-				sentence.MarkNewPhraseBase();

-				break;

-

-			case SPEI_HYPOTHESIS:

-			case SPEI_RECOGNITION:

-			case SPEI_FALSE_RECOGNITION:

-				{

-                    CComPtr<ISpRecoResult> cpResult;

-                    cpResult = spEvent.RecoResult();

-

-                    CSpDynamicString dstrText;

-                    if (spEvent.eEventId == SPEI_FALSE_RECOGNITION)

-                    {

-                        dstrText = L"(Unrecognized)";

-

-						result = SR_RESULT_FAILED;

-

-						// It's possible that the failed recog might have more words, so see if that's the case

-						EnumeratePhonemes( cpPhoneConv, cpResult, sentence );

-					}

-                    else

-                    {

-						// Hypothesis or recognition success

-                        cpResult->GetText( (ULONG)SP_GETWHOLEPHRASE, (ULONG)SP_GETWHOLEPHRASE, TRUE, &dstrText, NULL);

-

-						EnumeratePhonemes( cpPhoneConv, cpResult, sentence );

-

-						if ( spEvent.eEventId == SPEI_RECOGNITION )

-						{

-							result = SR_RESULT_SUCCESS;

-						}

-

-						pfnPrint( va( "%s%s\r\n", spEvent.eEventId == SPEI_HYPOTHESIS ? "[ Hypothesis ] " : "", dstrText.CopyToChar() ) );

-					}

-                    

-                    cpResult.Release();

-				}

-				break;

-				// end of the wav file was reached by the speech recognition engine

-            case SPEI_END_SR_STREAM:

-				fEndStreamReached = TRUE;

-				break;

-			}

-			

-			// clear any event data/object references

-			spEvent.Clear();

-		}// END event pulling loop - break on empty event queue OR end stream

-	}// END event polling loop - break on event timeout OR end stream

-	

-	// Deactivate rule

-	hr = cpRecoGrammar->SetRuleState( NULL, NULL, SPRS_INACTIVE );

-	if ( FAILED( hr ) )

-	{

-		pfnPrint( "Error:  SAPI 5.1 Unable to deactivate rule set\n" );

-		return result;

-	}

-

-	// close the input stream, since we're done with it

-	// NOTE: smart pointer will call SpStream's destructor, and consequently ::Close, but code may want to check for errors on ::Close operation

-	hr = cpInputStream->Close();

-	if ( FAILED( hr ) )

-	{

-		pfnPrint( "Error:  SAPI 5.1 Unable to close input stream\n" );

-		return result;

-	}

-

-	return result;

-}

-

-//-----------------------------------------------------------------------------

-// Purpose: HACK HACK:  We have to delete the RecoContext key or sapi starts to train

-//  itself on each iteration which was causing some problems.

-// Input  : hKey - 

-//-----------------------------------------------------------------------------

-void RecursiveRegDelKey(HKEY hKey)

-{

-	char keyname[256]={0};

-	DWORD namesize=256;

-

-	//base case: no subkeys when RegEnumKeyEx returns error on index 0

-	LONG lResult=RegEnumKeyEx(hKey,0,keyname,&namesize,NULL,NULL,NULL,NULL);

-	if (lResult!=ERROR_SUCCESS)

-	{

-		return;

-	}

-

-	do

-	{

-		HKEY subkey;

-		LONG lResult2;

-		LONG lDelResult;

-		lResult2=RegOpenKeyEx(hKey,keyname,0,KEY_ALL_ACCESS,&subkey);

-		

-		if (lResult2==ERROR_SUCCESS)

-		{

-			RecursiveRegDelKey(subkey);

-

-			RegCloseKey(subkey);

-			lDelResult=RegDeleteKey(hKey,keyname);

-			namesize=256;

-			//use 0 in the next function call because when you delete one, the rest shift down!

-			lResult=RegEnumKeyEx(hKey,0,keyname,&namesize,NULL,NULL,NULL,NULL);

-		}

-

-		else 

-		{

-			break;

-		}

-

-	} while (lResult!=ERROR_NO_MORE_ITEMS);

-}

-

-bool IsUseable( CWordTag *word )

-{

-	if ( word->m_uiStartByte || word->m_uiEndByte )

-		return true;

-

-	return false;

-}

-

-int FindLastUsableWord( CSentence& outwords )

-{

-	int numwords = outwords.m_Words.Size();

-	if ( numwords < 1 )

-	{

-		Assert( 0 );

-		return -1;

-	}

-

-	for ( int i = numwords-1; i >= 0; i-- )

-	{

-		CWordTag *check = outwords.m_Words[ i ];

-		if ( IsUseable( check ) )

-		{

-			return i;

-		}

-	}

-

-	return -1;

-}

-

-

-int FindFirstUsableWord( CSentence& outwords )

-{

-	int numwords = outwords.m_Words.Size();

-	if ( numwords < 1 )

-	{

-		Assert( 0 );

-		return -1;

-	}

-

-	for ( int i = 0; i < numwords; i++ )

-	{

-		CWordTag *check = outwords.m_Words[ i ];

-		if ( IsUseable( check ) )

-		{

-			return i;

-		}

-	}

-

-	return -1;

-}

-

-//-----------------------------------------------------------------------------

-// Purpose: Counts words which have either a valid start or end byte

-// Input  : *outwords - 

-// Output : int

-//-----------------------------------------------------------------------------

-int CountUsableWords( CSentence& outwords )

-{

-	int count = 0;

-	int numwords = outwords.m_Words.Size();

-	// Nothing to do

-	if ( numwords <= 0 )

-		return count;

-

-	for ( int i = 0; i < numwords; i++ )

-	{

-		CWordTag *word = outwords.m_Words[ i ];

-		if ( !IsUseable( word ) )

-			continue;

-

-		count++;

-	}

-

-	return count;

-}

-

-

-//-----------------------------------------------------------------------------

-// Purpose: Counts words which have either a valid start or end byte

-// Input  : *outwords - 

-// Output : int

-//-----------------------------------------------------------------------------

-int CountUnuseableWords( CSentence& outwords )

-{

-	int count = 0;

-	int numwords = outwords.m_Words.Size();

-	// Nothing to do

-	if ( numwords <= 0 )

-		return count;

-

-	for ( int i = 0; i < numwords; i++ )

-	{

-		CWordTag *word = outwords.m_Words[ i ];

-		if ( IsUseable( word ) )

-			continue;

-

-		count++;

-	}

-

-	return count;

-}

-

-// Keeps same relative spacing, but rebases list

-void RepartitionPhonemes( CWordTag *word, unsigned int oldStart, unsigned int oldEnd )

-{

-	// Repartition phonemes based on old range

-	float oldRange = ( float )( oldEnd - oldStart );

-	float newRange = ( float )( word->m_uiEndByte - word->m_uiStartByte );

-

-	for ( int i = 0; i < word->m_Phonemes.Size(); i++ )

-	{

-		CPhonemeTag *tag = word->m_Phonemes[ i ];

-		Assert( tag );

-

-		float frac1 = 0.0f, frac2 = 0.0f;

-		float delta1, delta2;

-		

-		delta1 = ( float ) ( tag->m_uiStartByte - oldStart );

-		delta2 = ( float ) ( tag->m_uiEndByte - oldStart );

-		if ( oldRange > 0.0f )

-		{

-			frac1 = delta1 / oldRange;

-			frac2 = delta2 / oldRange;

-		}

-

-		tag->m_uiStartByte = word->m_uiStartByte + ( unsigned int ) ( frac1 * newRange );

-		tag->m_uiEndByte = word->m_uiStartByte +  ( unsigned int ) ( frac2 * newRange );

-	}

-}

-

-void PartitionWords( CSentence& outwords, int start, int end, int sampleStart, int sampleEnd )

-{

-	int wordCount = end - start + 1;

-	Assert( wordCount >= 1 );

-	int stepSize  = ( sampleEnd - sampleStart ) / wordCount;

-

-	int currentStart = sampleStart;

-

-	for ( int i = start; i <= end; i++ )

-	{

-		CWordTag *word = outwords.m_Words[ i ];

-		Assert( word );

-

-		unsigned int oldStart = word->m_uiStartByte;

-		unsigned int oldEnd = word->m_uiEndByte;

-

-		word->m_uiStartByte = currentStart;

-		word->m_uiEndByte = currentStart + stepSize;

-

-		RepartitionPhonemes( word, oldStart, oldEnd );

-

-		currentStart += stepSize;

-	}

-}

-

-void MergeWords( CWordTag *w1, CWordTag *w2 )

-{

-	unsigned int start, end;

-

-	start = min( w1->m_uiStartByte, w2->m_uiStartByte );

-	end = max( w1->m_uiEndByte, w2->m_uiEndByte );

-

-	unsigned int mid = ( start + end ) / 2;

-

-	unsigned int oldw1start, oldw2start, oldw1end, oldw2end;

-

-	oldw1start = w1->m_uiStartByte;

-	oldw2start = w2->m_uiStartByte;

-	oldw1end = w1->m_uiEndByte;

-	oldw2end = w2->m_uiEndByte;

-

-	w1->m_uiStartByte = start;

-	w1->m_uiEndByte = mid;

-	w2->m_uiStartByte = mid;

-	w2->m_uiEndByte = end;

-

-	RepartitionPhonemes( w1, oldw1start, oldw1end );

-	RepartitionPhonemes( w2, oldw2start, oldw2end );

-}

-

-void FixupZeroLengthWords( CSentence& outwords )

-{

-	while ( 1 )

-	{

-		int i;

-		for ( i = 0 ; i < outwords.m_Words.Size() - 1; i++ )

-		{

-			CWordTag *current, *next;

-

-			current = outwords.m_Words[ i ];

-			next = outwords.m_Words[ i + 1 ];

-

-			if ( current->m_uiEndByte - current->m_uiStartByte <= 0 )

-			{

-				MergeWords( current, next );

-				break;

-			}

-

-			if ( next->m_uiEndByte - next->m_uiStartByte <= 0 )

-			{

-				MergeWords( current, next );

-				break;

-			}

-		}

-

-		if ( i >= outwords.m_Words.Size() - 1 )

-		{

-			break;

-		}

-	}

-}

-

-void ComputeMissingByteSpans( int numsamples, CSentence& outwords )

-{

-	int numwords = outwords.m_Words.Size();

-	// Nothing to do

-	if ( numwords <= 0 )

-		return;

-

-	int interationcount = 1;

-

-	while( 1 )

-	{

-		Log( "\nCompute %i\n", interationcount++ );

-		LogWords( outwords );

-

-		int wordNumber;

-

-		// Done!

-		if ( !CountUnuseableWords( outwords ) )

-		{

-			FixupZeroLengthWords( outwords );

-			break;

-		}

-

-		if ( !CountUsableWords( outwords ) )

-		{

-			// Evenly space words across full sample time

-			PartitionWords( outwords, 0, numwords - 1, 0, numsamples );

-			break;

-		}

-

-		wordNumber = FindFirstUsableWord( outwords );

-		// Not the first word

-		if ( wordNumber > 0 )

-		{

-			// Repartition all of the unusables and the first one starting at zero over the range

-			CWordTag *firstUsable = outwords.m_Words[ wordNumber ];

-			Assert( firstUsable );

-

-			if ( firstUsable->m_uiStartByte != 0 )

-			{

-				PartitionWords( outwords, 0, wordNumber - 1, 0, firstUsable->m_uiStartByte );

-			}

-			else

-			{

-				PartitionWords( outwords, 0, wordNumber, 0, firstUsable->m_uiEndByte );

-			}

-

-			// Start over

-			continue;

-		}

-

-		wordNumber = FindLastUsableWord( outwords );

-		// Not the last word

-		if ( wordNumber >= 0 && wordNumber < numwords - 1 )

-		{

-			// Repartition all of the unusables and the first one starting at zero over the range

-			CWordTag *lastUsable = outwords.m_Words[ wordNumber ];

-			Assert( lastUsable );

-

-			if ( lastUsable->m_uiEndByte != (unsigned int)numsamples )

-			{

-				PartitionWords( outwords, wordNumber + 1, numwords-1, lastUsable->m_uiEndByte, numsamples );

-			}

-			else

-			{

-				PartitionWords( outwords, wordNumber, numwords-1, lastUsable->m_uiStartByte, numsamples );

-			}

-

-			// Start over

-			continue;

-		}

-

-		// If we get here it means that the start and end of the list are okay and we just have to

-		//  iterate across the list and fix things in the middle

-		int startByte = 0;

-		int endByte = 0;

-		for ( int i = 0; i < numwords ; i++ )

-		{

-			CWordTag *word = outwords.m_Words[ i ];

-			if ( IsUseable( word ) )

-			{

-				startByte = word->m_uiEndByte;

-				continue;

-			}

-

-			// Found the start of a chain of 1 or more unusable words

-			// Find the startbyte of the next usable word and count how many words we check

-			int wordCount = 1;

-			for ( int j = i + 1; j < numwords; j++ )

-			{

-				CWordTag *next = outwords.m_Words[ j ];

-				if ( IsUseable( next ) )

-				{

-					endByte = next->m_uiStartByte;

-					break;

-				}

-

-				wordCount++;

-			}

-

-			// Now partition words across the gap and go to start again

-			PartitionWords( outwords, i, i + wordCount - 1, startByte, endByte );

-			break;

-		}

-	}

-}

-

-//-----------------------------------------------------------------------------

-// Purpose: Given a wavfile and a list of inwords, determines the word/phonene 

-//  sample counts for the sentce

-// Input  : *wavfile - 

-//			*inwords - 

-//			*outphonemes{	text.Clear( - 

-// Output : SR_RESULT

-//-----------------------------------------------------------------------------

-static SR_RESULT SAPI_ExtractPhonemes( 

-	const char *wavfile,

-	int numsamples,

-	void (*pfnPrint)( const char *fmt, ... ),

-	CSentence& inwords,

-	CSentence& outwords )

-{

-	LogReset();

-

-	USES_CONVERSION;

-

-	CSpDynamicString text;

-	text.Clear();

-

-	HKEY hkwipe;

-	LONG lResult = RegOpenKeyEx( HKEY_CURRENT_USER, "Software\\Microsoft\\Speech\\RecoProfiles", 0, KEY_ALL_ACCESS, &hkwipe );

-	if ( lResult == ERROR_SUCCESS )

-	{

-		RecursiveRegDelKey( hkwipe );

-		RegCloseKey( hkwipe );

-	}

-

-	if ( strlen( inwords.GetText() ) <= 0 )

-	{

-		inwords.SetTextFromWords();

-	}

-

-	// Construct a string from the inwords array

-	text.Append( T2W( inwords.GetText() ) );

-

-	// Assume failure

-	SR_RESULT result = SR_RESULT_ERROR;

-

-	if ( text.Length() > 0 )

-	{

-		CSentence sentence;

-

-		pfnPrint( "Processing...\r\n" );

-

-		// Give it a try

-		result = ExtractPhonemes( wavfile, text, sentence, pfnPrint );

-

-		pfnPrint( "Finished.\r\n" );

-		// PrintWordsAndPhonemes( sentence, pfnPrint );

-

-		// Copy results to outputs

-		outwords.Reset();

-

-		outwords.SetText( inwords.GetText() );

-		

-		Log( "Starting\n" );

-		LogWords( inwords );

-

-		if ( SR_RESULT_ERROR != result )

-		{

-			int i;

-

-			Log( "Hypothesized\n" );

-			LogWords( sentence );

-

-			for( i = 0 ; i < sentence.m_Words.Size(); i++ )

-			{

-				CWordTag *tag = sentence.m_Words[ i ];

-				if ( tag )

-				{

-					// Skip '...' tag

-					if ( stricmp( tag->GetWord(), "..." ) )

-					{

-						CWordTag *newTag = new CWordTag( *tag );

-

-						outwords.m_Words.AddToTail( newTag );

-					}

-				}

-			}

-

-			// Now insert unrecognized/skipped words from original list

-			//

-			int frompos = 0, topos = 0;

-

-			while( 1 )

-			{

-				// End of source list

-				if ( frompos >= inwords.m_Words.Size() )

-					break;

-

-				const CWordTag *fromTag = inwords.m_Words[ frompos ];

-

-				// Reached end of destination list, just copy words over from from source list until

-				//  we run out of source words

-				if ( topos >= outwords.m_Words.Size() )

-				{

-					// Just copy words over

-					CWordTag *newWord = new CWordTag( *fromTag );

-

-					// Remove phonemes

-					while ( newWord->m_Phonemes.Size() > 0 )

-					{

-						CPhonemeTag *kill = newWord->m_Phonemes[ 0 ];

-						newWord->m_Phonemes.Remove( 0 );

-						delete kill;

-					}

-

-					outwords.m_Words.AddToTail( newWord );

-					frompos++;

-					topos++;

-					continue;

-				}

-

-				// Destination word

-				const CWordTag *toTag = outwords.m_Words[ topos ];

-

-				// Words match, just skip ahead

-				if ( !stricmp( fromTag->GetWord(), toTag->GetWord() ) )

-				{

-					frompos++;

-					topos++;

-					continue;

-				}

-

-				// The only case we handle is that something in the source wasn't in the destination

-

-				// Find the next source word that appears in the destination

-				int skipAhead = frompos + 1;

-				bool found = false;

-				while ( skipAhead < inwords.m_Words.Size() )

-				{

-					const CWordTag *sourceWord = inwords.m_Words[ skipAhead ];

-					if ( !stricmp( sourceWord->GetWord(), toTag->GetWord() ) )

-					{

-						found = true;

-						break;

-					}

-

-					skipAhead++;

-				}

-

-				// Uh oh destination has words that are not in source, just skip to next destination word?

-				if ( !found )

-				{

-					topos++;

-				}

-				else

-				{

-					// Copy words from from source list into destination

-					// 

-					int skipCount = skipAhead - frompos;

-

-					while ( --skipCount>= 0 )

-					{

-						const CWordTag *sourceWord = inwords.m_Words[ frompos++ ];

-						CWordTag *newWord = new CWordTag( *sourceWord );

-

-						// Remove phonemes

-						while ( newWord->m_Phonemes.Size() > 0 )

-						{

-							CPhonemeTag *kill = newWord->m_Phonemes[ 0 ];

-							newWord->m_Phonemes.Remove( 0 );

-							delete kill;

-						}

-

-						outwords.m_Words.InsertBefore( topos, newWord );

-						topos++;

-					}

-

-					frompos++;

-					topos++;

-				}

-			}

-

-			Log( "\nDone simple check\n" );

-

-			LogWords( outwords );

-			LogPhonemes( outwords );

-

-			ComputeMissingByteSpans( numsamples, outwords );

-

-			Log( "\nFinal check\n" );

-

-			LogWords( outwords );

-			LogPhonemes( outwords );

-		}

-	}

-	else

-	{

-		pfnPrint( "Input sentence is empty!\n" );

-	}

-

-	// Return results

-	return result;

-}

-

-

-//-----------------------------------------------------------------------------

-// Purpose: Expose the interface

-//-----------------------------------------------------------------------------

-class CPhonemeExtractorSAPI : public IPhonemeExtractor

-{

-public:

-	virtual PE_APITYPE	GetAPIType() const

-	{

-		return SPEECH_API_SAPI;

-	}

-

-	// Used for menus, etc

-	virtual char const *GetName() const

-	{

-		return "MS SAPI 5.1";

-	}

-

-	SR_RESULT Extract( 

-		const char *wavfile,

-		int numsamples,

-		void (*pfnPrint)( const char *fmt, ... ),

-		CSentence& inwords,

-		CSentence& outwords )

-	{

-		return SAPI_ExtractPhonemes( wavfile, numsamples, pfnPrint, inwords, outwords );

-	}

-};

-

+//========= Copyright Valve Corporation, All rights reserved. ============//
+//
+// Purpose: 
+//
+// $NoKeywords: $
+//
+//=============================================================================//
+// extracephonemes.cpp : Defines the entry point for the console application.
+//
+#define PROTECTED_THINGS_DISABLE
+
+#include "tier0/wchartypes.h"
+#include <stdio.h>
+#include <windows.h>
+#include <tchar.h>
+#include "sphelper.h"
+#include "spddkhlp.h"
+// ATL Header Files
+#include <atlbase.h>
+// Face poser and util includes
+#include "utlvector.h"
+#include "phonemeextractor/PhonemeExtractor.h"
+#include "PhonemeConverter.h"
+#include "sentence.h"
+#include "tier0/dbg.h"
+#include "tier0/icommandline.h"
+#include "filesystem.h"
+
+// Extract phoneme grammar id
+#define EP_GRAM_ID			101
+// First rule of dynamic sentence rule set
+#define DYN_SENTENCERULE	102
+// # of milliseconds to allow for processing before timeout
+#define SR_WAVTIMEOUT		4000
+// Weight tag for rule to rule word/rule transitions
+#define CONFIDENCE_WEIGHT	0.0f
+
+//#define LOGGING		1
+#define LOGFILE		"c:\\fp.log"
+
+void LogReset( void )
+{
+#if LOGGING
+	FILE *fp = fopen( LOGFILE, "w" );
+	if ( fp )
+		fclose( fp );
+#endif
+}
+
+char *va( const char *fmt, ... );
+
+//-----------------------------------------------------------------------------
+// Purpose: 
+// Input  : *words - 
+//-----------------------------------------------------------------------------
+void LogWords( CSentence& sentence )
+{
+	Log( "Wordcount == %i\n", sentence.m_Words.Size() );
+
+	for ( int i = 0; i < sentence.m_Words.Size(); i++ )
+	{
+		const CWordTag *w = sentence.m_Words[ i ];
+		Log( "Word %s %u to %u\n", w->GetWord(), w->m_uiStartByte, w->m_uiEndByte );
+	}
+}
+
+//-----------------------------------------------------------------------------
+// Purpose: 
+// Input  : *phonemes - 
+//-----------------------------------------------------------------------------
+void LogPhonemes( CSentence& sentence )
+{
+	return;
+
+	Log( "Phonemecount == %i\n", sentence.CountPhonemes() );
+
+	for ( int i = 0; i < sentence.m_Words.Size(); i++ )
+	{
+		const CWordTag *w = sentence.m_Words[ i ];
+
+		for ( int j = 0; j < w->m_Phonemes.Size(); j++ )
+		{
+			const CPhonemeTag *p = w->m_Phonemes[ j ];
+			Log( "Phoneme %s %u to %u\n", p->GetTag(), p->m_uiStartByte, p->m_uiEndByte );
+		}
+	}
+}
+
+#define NANO_CONVERT 10000000.0f;
+
+//-----------------------------------------------------------------------------
+// Purpose: Walk list of words and phonemes and create phoneme tags in CSentence object
+//  FIXME:  Right now, phonemes are assumed to evenly space out across a word.
+// Input  : *converter - 
+//			result - 
+//			sentence - 
+//-----------------------------------------------------------------------------
+void EnumeratePhonemes( ISpPhoneConverter *converter, const ISpRecoResult* result, CSentence& sentence )
+{
+	USES_CONVERSION;
+
+	// Grab access to element container
+	ISpPhrase *phrase = ( ISpPhrase * )result;
+	if ( !phrase )
+		return;
+
+    SPPHRASE *pElements;
+	if ( !SUCCEEDED( phrase->GetPhrase( &pElements ) ) )
+		return;
+
+	// Only use it if it's better/same size as what we already had on-hand
+	if ( pElements->Rule.ulCountOfElements > 0 )
+		//(unsigned int)( sentence.m_Words.Size() - sentence.GetWordBase() ) )
+	{
+		sentence.ResetToBase();
+
+		// Walk list of words
+		for ( ULONG i = 0; i < pElements->Rule.ulCountOfElements; i++ )
+		{
+			unsigned int wordstart, wordend;
+
+			// Get start/end sample index
+			wordstart	= pElements->pElements[i].ulAudioStreamOffset + (unsigned int)pElements->ullAudioStreamPosition;
+			wordend		= wordstart + pElements->pElements[i].ulAudioSizeBytes;
+
+			// Create word tag
+			CWordTag *w = new CWordTag( W2T( pElements->pElements[i].pszDisplayText ) );
+			Assert( w );
+			w->m_uiStartByte = wordstart;
+			w->m_uiEndByte   = wordend;
+
+			sentence.AddWordTag( w );
+
+			// Count # of phonemes in this word
+			SPPHONEID pstr[ 2 ];
+			pstr[ 1 ] = 0;
+			WCHAR wszPhoneme[ SP_MAX_PRON_LENGTH ];
+
+			const SPPHONEID *current;
+			SPPHONEID phoneme;
+			current = pElements->pElements[i].pszPronunciation;
+			float total_weight = 0.0f;
+			while ( 1 )
+			{
+				phoneme = *current++;
+				if ( !phoneme )
+					break;
+
+				pstr[ 0 ] = phoneme;
+				wszPhoneme[ 0 ] = L'\0';
+
+				converter->IdToPhone( pstr, wszPhoneme );
+
+				total_weight += WeightForPhoneme( W2A( wszPhoneme ) );
+			}
+
+			current = pElements->pElements[i].pszPronunciation;
+
+			// Decide # of bytes/phoneme weight
+			float psize = 0;
+			if ( total_weight )
+			{
+				psize = ( wordend - wordstart ) / total_weight;
+			}
+
+			int number = 0;
+
+			// Re-walk the phoneme list and create true phoneme tags
+			float startWeight = 0.0f;
+			while ( 1 )
+			{
+				phoneme = *current++;
+				if ( !phoneme )
+					break;
+
+				pstr[ 0 ] = phoneme;
+				wszPhoneme[ 0 ] = L'\0';
+
+				converter->IdToPhone( pstr, wszPhoneme );
+ 
+				CPhonemeTag *p = new CPhonemeTag( W2A( wszPhoneme ) );
+				Assert( p );
+				
+				float weight = WeightForPhoneme( W2A( wszPhoneme ) );
+
+				p->m_uiStartByte = wordstart + (int)( startWeight * psize );
+				p->m_uiEndByte	 = p->m_uiStartByte + (int)( psize * weight );
+
+				startWeight += weight;
+
+				// Convert to IPA phoneme code
+				p->SetPhonemeCode( TextToPhoneme( p->GetTag() ) );
+
+				sentence.AddPhonemeTag( w, p );
+
+				number++;
+			}
+		}	
+	}
+
+	// Free memory
+    ::CoTaskMemFree(pElements);
+}
+
+//-----------------------------------------------------------------------------
+// Purpose: Create rules for each word in the reference sentence
+//-----------------------------------------------------------------------------
+typedef struct
+{
+	int					ruleId;
+	SPSTATEHANDLE		hRule;
+	CSpDynamicString	word;
+	char				plaintext[ 256 ];
+} WORDRULETYPE;
+
+//-----------------------------------------------------------------------------
+// Purpose: Creates start for word of sentence
+// Input  : cpRecoGrammar - 
+//			*root - 
+//			*rules - 
+//			word - 
+//-----------------------------------------------------------------------------
+void AddWordRule( ISpRecoGrammar* cpRecoGrammar, SPSTATEHANDLE *root, CUtlVector< WORDRULETYPE > *rules, CSpDynamicString& word )
+{
+	USES_CONVERSION;
+	HRESULT hr;
+	WORDRULETYPE *newrule;
+
+	int idx = (*rules).AddToTail();
+
+	newrule = &(*rules)[ idx ];
+
+	newrule->ruleId = DYN_SENTENCERULE + idx + 1;
+	newrule->word = word;
+
+	strcpy( newrule->plaintext, W2T( word ) );
+
+	// Create empty rule
+	hr = cpRecoGrammar->CreateNewState( *root, &newrule->hRule );
+	Assert( !FAILED( hr ) );
+}
+
+//-----------------------------------------------------------------------------
+// Purpose: 
+// Input  : cpRecoGrammar - 
+//			*from - 
+//			*to - 
+//-----------------------------------------------------------------------------
+void AddWordTransitionRule( ISpRecoGrammar* cpRecoGrammar, WORDRULETYPE *from, WORDRULETYPE *to )
+{
+	USES_CONVERSION;
+
+	HRESULT hr;
+	Assert( from );
+
+	if ( from && !to )
+	{
+		OutputDebugString( va( "Transition from %s to TERM\r\n", from->plaintext ) );
+	}
+	else
+	{
+		OutputDebugString( va( "Transition from %s to %s\r\n", from->plaintext, to->plaintext ) );
+	}
+
+	hr = cpRecoGrammar->AddWordTransition( from->hRule, to ? to->hRule : NULL, (WCHAR *)from->word, NULL, SPWT_LEXICAL, CONFIDENCE_WEIGHT, NULL );
+	Assert( !FAILED( hr ) );
+}
+
+//-----------------------------------------------------------------------------
+// Purpose: 
+// Input  : cpRecoGrammar - 
+//			*from - 
+//			*to - 
+//-----------------------------------------------------------------------------
+void AddOptionalTransitionRule( ISpRecoGrammar* cpRecoGrammar, WORDRULETYPE *from, WORDRULETYPE *to )
+{
+	USES_CONVERSION;
+
+	HRESULT hr;
+	Assert( from );
+
+	if ( from && !to )
+	{
+		OutputDebugString( va( "Opt transition from %s to TERM\r\n", from->plaintext ) );
+	}
+	else
+	{
+		OutputDebugString( va( "Opt transition from %s to %s\r\n", from->plaintext, to->plaintext ) );
+	}
+
+	hr = cpRecoGrammar->AddWordTransition( from->hRule, to ? to->hRule : NULL, NULL, NULL, SPWT_LEXICAL, CONFIDENCE_WEIGHT, NULL );
+	Assert( !FAILED( hr ) );
+}
+
+#define MAX_WORD_SKIP 1
+//-----------------------------------------------------------------------------
+// Purpose: Links together all word rule states into a sentence rule CFG
+// Input  : singleword - 
+//			cpRecoGrammar - 
+//			*root - 
+//			*rules - 
+//-----------------------------------------------------------------------------
+bool BuildRules( ISpRecoGrammar* cpRecoGrammar, SPSTATEHANDLE *root, CUtlVector< WORDRULETYPE > *rules )
+{
+	HRESULT hr;
+	WORDRULETYPE *rule, *next;
+
+	int numrules = (*rules).Size();
+
+	rule = &(*rules)[ 0 ];
+
+	// Add transition
+	hr = cpRecoGrammar->AddWordTransition( *root, rule->hRule, NULL, NULL, SPWT_LEXICAL, CONFIDENCE_WEIGHT, NULL );
+	Assert( !FAILED( hr ) );
+
+	for ( int i = 0; i < numrules; i++ )
+	{
+		rule = &(*rules)[ i ];
+		if ( i < numrules - 1 )
+		{
+			next = &(*rules)[ i + 1 ];
+		}
+		else
+		{
+			next = NULL;
+		}
+
+		AddWordTransitionRule( cpRecoGrammar, rule, next );
+	}
+
+	if ( numrules > 1 )
+	{
+		for ( int skip = 1; skip <= min( MAX_WORD_SKIP, numrules ); skip++ )
+		{
+			OutputDebugString( va( "Opt transition from Root to %s\r\n", (*rules)[ 0 ].plaintext ) );
+
+			hr = cpRecoGrammar->AddWordTransition( *root, (*rules)[ 0 ].hRule, NULL, NULL, SPWT_LEXICAL, CONFIDENCE_WEIGHT, NULL );
+
+			// Now build rules where you can skip 1 to N intervening words
+			for ( int i = 1; i < numrules; i++ )
+			{
+				// Start at the beginning?
+				rule = &(*rules)[ i ];	
+				if ( i < numrules - skip )
+				{
+					next = &(*rules)[ i + skip ];
+				}
+				else
+				{
+					continue;
+				}
+
+				// Add transition
+				AddOptionalTransitionRule( cpRecoGrammar, rule, next );
+			}
+
+			// Go from final rule to end point
+			AddOptionalTransitionRule( cpRecoGrammar, rule, NULL );
+		}
+	}
+
+	// Store it
+	hr = cpRecoGrammar->Commit(NULL);
+	if ( FAILED( hr ) )
+		return false;
+
+	return true;
+}
+
+//-----------------------------------------------------------------------------
+// Purpose: Debugging, prints alternate list if one is created
+// Input  : cpResult - 
+//			(*pfnPrint - 
+//-----------------------------------------------------------------------------
+void PrintAlternates( ISpRecoResult* cpResult, void (*pfnPrint)( const char *fmt, ... ) )
+{
+	ISpPhraseAlt *rgPhraseAlt[ 32 ];
+	memset( rgPhraseAlt, 0, sizeof( rgPhraseAlt ) );
+
+	ULONG ulCount;
+	
+	ISpPhrase *phrase = ( ISpPhrase * )cpResult;
+	if ( phrase )
+	{
+		SPPHRASE *pElements;
+		if ( SUCCEEDED( phrase->GetPhrase( &pElements ) ) )
+		{
+			if ( pElements->Rule.ulCountOfElements > 0 )
+			{
+				HRESULT hr = cpResult->GetAlternates(
+					pElements->Rule.ulFirstElement,
+					pElements->Rule.ulCountOfElements, 
+					32,
+					rgPhraseAlt,
+					&ulCount);
+				
+				Assert( !FAILED( hr ) );
+				
+				for ( ULONG r = 0 ; r < ulCount; r++ )
+				{
+					CSpDynamicString dstrText;
+					hr = rgPhraseAlt[ r ]->GetText( (ULONG)SP_GETWHOLEPHRASE, (ULONG)SP_GETWHOLEPHRASE, TRUE, &dstrText, NULL);
+					Assert( !FAILED( hr ) );
+
+					pfnPrint( "[ ALT ]" );
+					pfnPrint( dstrText.CopyToChar() );
+					pfnPrint( "\r\n" );
+				}
+			}
+		}
+		
+	}
+
+	for ( int i = 0; i < 32; i++ )
+	{
+		if ( rgPhraseAlt[ i ] )
+		{
+			rgPhraseAlt[ i ]->Release();
+			rgPhraseAlt[ i ] = NULL;
+		}
+	}
+}
+
+void PrintWordsAndPhonemes( CSentence& sentence, void (*pfnPrint)( const char *fmt, ... ) )
+{
+	char sz[ 256 ];
+	int i;
+
+	pfnPrint( "WORDS\r\n\r\n" );
+
+	for ( i = 0 ; i < sentence.m_Words.Size(); i++ )
+	{
+		CWordTag *word = sentence.m_Words[ i ];
+		if ( !word )
+			continue;
+
+		sprintf( sz, "<%u - %u> %s\r\n", 
+			word->m_uiStartByte, word->m_uiEndByte, word->GetWord() );
+
+		pfnPrint( sz );
+
+		for ( int j = 0 ; j < word->m_Phonemes.Size(); j++ )
+		{
+			CPhonemeTag *phoneme = word->m_Phonemes[ j ];
+			if ( !phoneme )
+				continue;
+
+			sprintf( sz, "  <%u - %u> %s\r\n", 
+				phoneme->m_uiStartByte, phoneme->m_uiEndByte, phoneme->GetTag() );
+
+			pfnPrint( sz );
+		}
+	}
+
+	pfnPrint( "\r\n" );
+}
+
+//-----------------------------------------------------------------------------
+// Purpose: Given a wave file and a string of words "text", creates a CFG from the
+//  sentence and stores the resulting words/phonemes in CSentence
+// Input  : *wavname - 
+//			text - 
+//			sentence - 
+//			(*pfnPrint - 
+// Output : SR_RESULT
+//-----------------------------------------------------------------------------
+SR_RESULT ExtractPhonemes( const char *wavname, CSpDynamicString& text, CSentence& sentence, void (*pfnPrint)( const char *fmt, ...) )
+{
+	// Assume failure
+	SR_RESULT result = SR_RESULT_ERROR;
+
+	if ( text.Length() <= 0 )
+	{
+		pfnPrint( "Error:  no rule / text specified\n" );
+		return result;
+	}
+
+	USES_CONVERSION;
+	HRESULT hr;
+	
+	CUtlVector < WORDRULETYPE > wordRules;
+
+	CComPtr<ISpStream> cpInputStream;
+	CComPtr<ISpRecognizer> cpRecognizer;
+	CComPtr<ISpRecoContext> cpRecoContext;
+	CComPtr<ISpRecoGrammar> cpRecoGrammar;
+	CComPtr<ISpPhoneConverter>  cpPhoneConv;
+    
+	// Create basic SAPI stream object
+	// NOTE: The helper SpBindToFile can be used to perform the following operations
+	hr = cpInputStream.CoCreateInstance(CLSID_SpStream);
+	if ( FAILED( hr ) )
+	{
+		pfnPrint( "Error:  SAPI 5.1 Stream object not installed?\n" );
+		return result;
+	}
+
+	CSpStreamFormat sInputFormat;
+	
+	// setup stream object with wav file MY_WAVE_AUDIO_FILENAME
+	//   for read-only access, since it will only be access by the SR engine
+	hr = cpInputStream->BindToFile(
+		T2W(wavname),
+		SPFM_OPEN_READONLY,
+		NULL,
+		sInputFormat.WaveFormatExPtr(),
+		SPFEI_ALL_EVENTS );
+
+	if ( FAILED( hr ) )
+	{
+		pfnPrint( "Error: couldn't open wav file %s\n", wavname );
+		return result;
+	}
+	
+	// Create in-process speech recognition engine
+	hr = cpRecognizer.CoCreateInstance(CLSID_SpInprocRecognizer);
+	if ( FAILED( hr ) )
+	{
+		pfnPrint( "Error:  SAPI 5.1 In process recognizer object not installed?\n" );
+		return result;
+	}
+
+	// Create recognition context to receive events
+	hr = cpRecognizer->CreateRecoContext(&cpRecoContext);
+	if ( FAILED( hr ) )
+	{
+		pfnPrint( "Error:  SAPI 5.1 Unable to create recognizer context\n" );
+		return result;
+	}
+	
+	// Create a grammar
+	hr = cpRecoContext->CreateGrammar( EP_GRAM_ID, &cpRecoGrammar );
+	if ( FAILED( hr ) )
+	{
+		pfnPrint( "Error:  SAPI 5.1 Unable to create recognizer grammar\n" );
+		return result;
+	}
+
+	LANGID englishID = 0x409; // 1033 decimal
+
+	bool userSpecified = false;
+	LANGID langID = SpGetUserDefaultUILanguage();
+
+	// Allow commandline override
+	if ( CommandLine()->FindParm( "-languageid" ) != 0 )
+	{
+		userSpecified = true;
+		langID = CommandLine()->ParmValue( "-languageid", langID );
+	}
+
+	// Create a phoneme converter ( so we can convert to IPA codes )
+	hr = SpCreatePhoneConverter( langID, NULL, NULL, &cpPhoneConv );
+	if ( FAILED( hr ) )
+	{
+		if ( langID != englishID )
+		{
+			if ( userSpecified )
+			{
+				pfnPrint( "Warning:  SAPI 5.1 Unable to create phoneme converter for command line override -languageid %i\n", langID );
+			}
+			else
+			{
+				pfnPrint( "Warning:  SAPI 5.1 Unable to create phoneme converter for default UI language %i\n",langID );
+			}
+
+			// Try english!!!
+			langID = englishID;
+			hr = SpCreatePhoneConverter( langID, NULL, NULL, &cpPhoneConv );
+		}
+
+		if ( FAILED( hr ) )
+		{
+			pfnPrint( "Error:  SAPI 5.1 Unable to create phoneme converter for English language id %i\n", langID );
+			return result;
+		}
+		else
+		{
+			pfnPrint( "Note:  SAPI 5.1 Falling back to use english -languageid %i\n", langID );
+		}
+	}
+	else if ( userSpecified )
+	{
+		pfnPrint( "Note:  SAPI 5.1 Using user specified -languageid %i\n",langID );
+	}
+
+	SPSTATEHANDLE hStateRoot;
+	// create/re-create Root level rule of grammar
+	hr = cpRecoGrammar->GetRule(L"Root", 0, SPRAF_TopLevel | SPRAF_Active, TRUE, &hStateRoot);
+	if ( FAILED( hr ) )
+	{
+		pfnPrint( "Error:  SAPI 5.1 Unable to create root rule\n" );
+		return result;
+	}
+
+	// Inactivate it so we can alter it
+	hr = cpRecoGrammar->SetRuleState( NULL, NULL, SPRS_INACTIVE );
+	if ( FAILED( hr ) )
+	{
+		pfnPrint( "Error:  SAPI 5.1 Unable to deactivate grammar rules\n" );
+		return result;
+	}
+
+	// Create the rule set from the words in text
+	{
+		CSpDynamicString currentWord;
+		WCHAR *pos = ( WCHAR * )text;
+		WCHAR str[ 2 ];
+		str[1]= 0;
+
+		while ( *pos )
+		{
+			if ( *pos == L' ' /*|| *pos == L'.' || *pos == L'-'*/ )
+			{
+				// Add word to rule set
+				if ( currentWord.Length() > 0 )
+				{
+					AddWordRule( cpRecoGrammar, &hStateRoot, &wordRules, currentWord );
+					currentWord.Clear();
+				}
+				pos++;
+				continue;
+			}
+
+			// Skip anything that's inside a [ xxx ] pair.
+			if ( *pos == L'[' )
+			{
+				while ( *pos && *pos != L']' )
+				{
+					pos++;
+				}
+
+				if ( *pos )
+				{
+					pos++;
+				}
+				continue;
+			}
+
+			str[ 0 ] = *pos;
+
+			currentWord.Append( str );
+			pos++;
+		}
+
+		if ( currentWord.Length() > 0 )
+		{
+			AddWordRule( cpRecoGrammar, &hStateRoot, &wordRules, currentWord );
+		}
+
+		if ( wordRules.Size() <= 0 )
+		{
+			pfnPrint( "Error:  Text %s contained no usable words\n", text );
+			return result;
+		}
+
+		// Build all word to word transitions in the grammar
+		if ( !BuildRules( cpRecoGrammar, &hStateRoot, &wordRules ) )
+		{
+			pfnPrint( "Error:  Rule set for %s could not be generated\n", text );
+			return result;
+		}
+	}
+
+	// check for recognitions and end of stream event
+	const ULONGLONG ullInterest = 
+		SPFEI(SPEI_RECOGNITION) | SPFEI(SPEI_END_SR_STREAM) | SPFEI(SPEI_FALSE_RECOGNITION) | 
+		SPFEI(SPEI_PHRASE_START ) | SPFEI(SPEI_HYPOTHESIS ) | SPFEI(SPEI_INTERFERENCE) ;
+	hr = cpRecoContext->SetInterest( ullInterest, ullInterest );
+	if ( FAILED( hr ) )
+	{
+		pfnPrint( "Error:  SAPI 5.1 Unable to set interest level\n" );
+		return result;
+	}	
+	// use Win32 events for command-line style application
+	hr = cpRecoContext->SetNotifyWin32Event();
+	if ( FAILED( hr ) )
+	{
+		pfnPrint( "Error:  SAPI 5.1 Unable to set win32 notify event\n" );
+		return result;
+	}
+	// connect wav input to recognizer
+	// SAPI will negotiate mismatched engine/input audio formats using system audio codecs, so second parameter is not important - use default of TRUE
+	hr = cpRecognizer->SetInput(cpInputStream, TRUE);
+	if ( FAILED( hr ) )
+	{
+		pfnPrint( "Error:  SAPI 5.1 Unable to associate input stream\n" );
+		return result;
+	}	
+
+	// Activate the CFG ( rather than using dictation )
+	hr = cpRecoGrammar->SetRuleState( NULL, NULL, SPRS_ACTIVE );
+	if ( FAILED( hr ) )
+	{
+		switch ( hr )
+		{
+		case E_INVALIDARG:
+			pfnPrint( "pszName is invalid or bad. Alternatively, pReserved is non-NULL\n" );
+			break;
+		case SP_STREAM_UNINITIALIZED:
+			pfnPrint( "ISpRecognizer::SetInput has not been called with the InProc recognizer\n" );
+			break;
+		case SPERR_UNINITIALIZED:
+			pfnPrint( "The object has not been properly initialized.\n");
+			break;
+		case SPERR_UNSUPPORTED_FORMAT:
+			pfnPrint( "Audio format is bad or is not recognized. Alternatively, the device driver may be busy by another application and cannot be accessed.\n" );
+			break;
+		case SPERR_NOT_TOPLEVEL_RULE:
+			pfnPrint( "The rule pszName exists, but is not a top-level rule.\n" );
+			break;
+		default:
+			pfnPrint( "Unknown error\n" );
+			break;
+		}
+		pfnPrint( "Error:  SAPI 5.1 Unable to activate rule set\n" );
+		return result;
+	}
+
+	// while events occur, continue processing
+	// timeout should be greater than the audio stream length, or a reasonable amount of time expected to pass before no more recognitions are expected in an audio stream
+	BOOL fEndStreamReached = FALSE;
+	while (!fEndStreamReached && S_OK == cpRecoContext->WaitForNotifyEvent( SR_WAVTIMEOUT ))
+	{
+		CSpEvent spEvent;
+		// pull all queued events from the reco context's event queue
+		
+		while (!fEndStreamReached && S_OK == spEvent.GetFrom(cpRecoContext))
+		{
+			// Check event type
+			switch (spEvent.eEventId)
+			{
+			case SPEI_INTERFERENCE:
+				{
+					SPINTERFERENCE interference = spEvent.Interference();
+
+					switch ( interference )
+					{
+					case SPINTERFERENCE_NONE:
+						pfnPrint( "[ I None ]\r\n" );
+						break;
+					case SPINTERFERENCE_NOISE:
+						pfnPrint( "[ I Noise ]\r\n" );
+						break;
+					case SPINTERFERENCE_NOSIGNAL:
+						pfnPrint( "[ I No Signal ]\r\n" );
+						break;
+					case SPINTERFERENCE_TOOLOUD:
+						pfnPrint( "[ I Too Loud ]\r\n" );
+						break;
+					case SPINTERFERENCE_TOOQUIET:
+						pfnPrint( "[ I Too Quiet ]\r\n" );
+						break;
+					case SPINTERFERENCE_TOOFAST:
+						pfnPrint( "[ I Too Fast ]\r\n" );
+						break;
+					case SPINTERFERENCE_TOOSLOW:
+						pfnPrint( "[ I Too Slow ]\r\n" );
+						break;
+					default:
+						break;
+					}
+				}
+				break;
+			case SPEI_PHRASE_START:
+				pfnPrint( "Phrase Start\r\n" );
+				sentence.MarkNewPhraseBase();
+				break;
+
+			case SPEI_HYPOTHESIS:
+			case SPEI_RECOGNITION:
+			case SPEI_FALSE_RECOGNITION:
+				{
+                    CComPtr<ISpRecoResult> cpResult;
+                    cpResult = spEvent.RecoResult();
+
+                    CSpDynamicString dstrText;
+                    if (spEvent.eEventId == SPEI_FALSE_RECOGNITION)
+                    {
+                        dstrText = L"(Unrecognized)";
+
+						result = SR_RESULT_FAILED;
+
+						// It's possible that the failed recog might have more words, so see if that's the case
+						EnumeratePhonemes( cpPhoneConv, cpResult, sentence );
+					}
+                    else
+                    {
+						// Hypothesis or recognition success
+                        cpResult->GetText( (ULONG)SP_GETWHOLEPHRASE, (ULONG)SP_GETWHOLEPHRASE, TRUE, &dstrText, NULL);
+
+						EnumeratePhonemes( cpPhoneConv, cpResult, sentence );
+
+						if ( spEvent.eEventId == SPEI_RECOGNITION )
+						{
+							result = SR_RESULT_SUCCESS;
+						}
+
+						pfnPrint( va( "%s%s\r\n", spEvent.eEventId == SPEI_HYPOTHESIS ? "[ Hypothesis ] " : "", dstrText.CopyToChar() ) );
+					}
+                    
+                    cpResult.Release();
+				}
+				break;
+				// end of the wav file was reached by the speech recognition engine
+            case SPEI_END_SR_STREAM:
+				fEndStreamReached = TRUE;
+				break;
+			}
+			
+			// clear any event data/object references
+			spEvent.Clear();
+		}// END event pulling loop - break on empty event queue OR end stream
+	}// END event polling loop - break on event timeout OR end stream
+	
+	// Deactivate rule
+	hr = cpRecoGrammar->SetRuleState( NULL, NULL, SPRS_INACTIVE );
+	if ( FAILED( hr ) )
+	{
+		pfnPrint( "Error:  SAPI 5.1 Unable to deactivate rule set\n" );
+		return result;
+	}
+
+	// close the input stream, since we're done with it
+	// NOTE: smart pointer will call SpStream's destructor, and consequently ::Close, but code may want to check for errors on ::Close operation
+	hr = cpInputStream->Close();
+	if ( FAILED( hr ) )
+	{
+		pfnPrint( "Error:  SAPI 5.1 Unable to close input stream\n" );
+		return result;
+	}
+
+	return result;
+}
+
+//-----------------------------------------------------------------------------
+// Purpose: HACK HACK:  We have to delete the RecoContext key or sapi starts to train
+//  itself on each iteration which was causing some problems.
+// Input  : hKey - 
+//-----------------------------------------------------------------------------
+void RecursiveRegDelKey(HKEY hKey)
+{
+	char keyname[256]={0};
+	DWORD namesize=256;
+
+	//base case: no subkeys when RegEnumKeyEx returns error on index 0
+	LONG lResult=RegEnumKeyEx(hKey,0,keyname,&namesize,NULL,NULL,NULL,NULL);
+	if (lResult!=ERROR_SUCCESS)
+	{
+		return;
+	}
+
+	do
+	{
+		HKEY subkey;
+		LONG lResult2;
+		LONG lDelResult;
+		lResult2=RegOpenKeyEx(hKey,keyname,0,KEY_ALL_ACCESS,&subkey);
+		
+		if (lResult2==ERROR_SUCCESS)
+		{
+			RecursiveRegDelKey(subkey);
+
+			RegCloseKey(subkey);
+			lDelResult=RegDeleteKey(hKey,keyname);
+			namesize=256;
+			//use 0 in the next function call because when you delete one, the rest shift down!
+			lResult=RegEnumKeyEx(hKey,0,keyname,&namesize,NULL,NULL,NULL,NULL);
+		}
+
+		else 
+		{
+			break;
+		}
+
+	} while (lResult!=ERROR_NO_MORE_ITEMS);
+}
+
+bool IsUseable( CWordTag *word )
+{
+	if ( word->m_uiStartByte || word->m_uiEndByte )
+		return true;
+
+	return false;
+}
+
+int FindLastUsableWord( CSentence& outwords )
+{
+	int numwords = outwords.m_Words.Size();
+	if ( numwords < 1 )
+	{
+		Assert( 0 );
+		return -1;
+	}
+
+	for ( int i = numwords-1; i >= 0; i-- )
+	{
+		CWordTag *check = outwords.m_Words[ i ];
+		if ( IsUseable( check ) )
+		{
+			return i;
+		}
+	}
+
+	return -1;
+}
+
+
+int FindFirstUsableWord( CSentence& outwords )
+{
+	int numwords = outwords.m_Words.Size();
+	if ( numwords < 1 )
+	{
+		Assert( 0 );
+		return -1;
+	}
+
+	for ( int i = 0; i < numwords; i++ )
+	{
+		CWordTag *check = outwords.m_Words[ i ];
+		if ( IsUseable( check ) )
+		{
+			return i;
+		}
+	}
+
+	return -1;
+}
+
+//-----------------------------------------------------------------------------
+// Purpose: Counts words which have either a valid start or end byte
+// Input  : *outwords - 
+// Output : int
+//-----------------------------------------------------------------------------
+int CountUsableWords( CSentence& outwords )
+{
+	int count = 0;
+	int numwords = outwords.m_Words.Size();
+	// Nothing to do
+	if ( numwords <= 0 )
+		return count;
+
+	for ( int i = 0; i < numwords; i++ )
+	{
+		CWordTag *word = outwords.m_Words[ i ];
+		if ( !IsUseable( word ) )
+			continue;
+
+		count++;
+	}
+
+	return count;
+}
+
+
+//-----------------------------------------------------------------------------
+// Purpose: Counts words which have either a valid start or end byte
+// Input  : *outwords - 
+// Output : int
+//-----------------------------------------------------------------------------
+int CountUnuseableWords( CSentence& outwords )
+{
+	int count = 0;
+	int numwords = outwords.m_Words.Size();
+	// Nothing to do
+	if ( numwords <= 0 )
+		return count;
+
+	for ( int i = 0; i < numwords; i++ )
+	{
+		CWordTag *word = outwords.m_Words[ i ];
+		if ( IsUseable( word ) )
+			continue;
+
+		count++;
+	}
+
+	return count;
+}
+
+// Keeps same relative spacing, but rebases list
+void RepartitionPhonemes( CWordTag *word, unsigned int oldStart, unsigned int oldEnd )
+{
+	// Repartition phonemes based on old range
+	float oldRange = ( float )( oldEnd - oldStart );
+	float newRange = ( float )( word->m_uiEndByte - word->m_uiStartByte );
+
+	for ( int i = 0; i < word->m_Phonemes.Size(); i++ )
+	{
+		CPhonemeTag *tag = word->m_Phonemes[ i ];
+		Assert( tag );
+
+		float frac1 = 0.0f, frac2 = 0.0f;
+		float delta1, delta2;
+		
+		delta1 = ( float ) ( tag->m_uiStartByte - oldStart );
+		delta2 = ( float ) ( tag->m_uiEndByte - oldStart );
+		if ( oldRange > 0.0f )
+		{
+			frac1 = delta1 / oldRange;
+			frac2 = delta2 / oldRange;
+		}
+
+		tag->m_uiStartByte = word->m_uiStartByte + ( unsigned int ) ( frac1 * newRange );
+		tag->m_uiEndByte = word->m_uiStartByte +  ( unsigned int ) ( frac2 * newRange );
+	}
+}
+
+void PartitionWords( CSentence& outwords, int start, int end, int sampleStart, int sampleEnd )
+{
+	int wordCount = end - start + 1;
+	Assert( wordCount >= 1 );
+	int stepSize  = ( sampleEnd - sampleStart ) / wordCount;
+
+	int currentStart = sampleStart;
+
+	for ( int i = start; i <= end; i++ )
+	{
+		CWordTag *word = outwords.m_Words[ i ];
+		Assert( word );
+
+		unsigned int oldStart = word->m_uiStartByte;
+		unsigned int oldEnd = word->m_uiEndByte;
+
+		word->m_uiStartByte = currentStart;
+		word->m_uiEndByte = currentStart + stepSize;
+
+		RepartitionPhonemes( word, oldStart, oldEnd );
+
+		currentStart += stepSize;
+	}
+}
+
+void MergeWords( CWordTag *w1, CWordTag *w2 )
+{
+	unsigned int start, end;
+
+	start = min( w1->m_uiStartByte, w2->m_uiStartByte );
+	end = max( w1->m_uiEndByte, w2->m_uiEndByte );
+
+	unsigned int mid = ( start + end ) / 2;
+
+	unsigned int oldw1start, oldw2start, oldw1end, oldw2end;
+
+	oldw1start = w1->m_uiStartByte;
+	oldw2start = w2->m_uiStartByte;
+	oldw1end = w1->m_uiEndByte;
+	oldw2end = w2->m_uiEndByte;
+
+	w1->m_uiStartByte = start;
+	w1->m_uiEndByte = mid;
+	w2->m_uiStartByte = mid;
+	w2->m_uiEndByte = end;
+
+	RepartitionPhonemes( w1, oldw1start, oldw1end );
+	RepartitionPhonemes( w2, oldw2start, oldw2end );
+}
+
+void FixupZeroLengthWords( CSentence& outwords )
+{
+	while ( 1 )
+	{
+		int i;
+		for ( i = 0 ; i < outwords.m_Words.Size() - 1; i++ )
+		{
+			CWordTag *current, *next;
+
+			current = outwords.m_Words[ i ];
+			next = outwords.m_Words[ i + 1 ];
+
+			if ( current->m_uiEndByte - current->m_uiStartByte <= 0 )
+			{
+				MergeWords( current, next );
+				break;
+			}
+
+			if ( next->m_uiEndByte - next->m_uiStartByte <= 0 )
+			{
+				MergeWords( current, next );
+				break;
+			}
+		}
+
+		if ( i >= outwords.m_Words.Size() - 1 )
+		{
+			break;
+		}
+	}
+}
+
+void ComputeMissingByteSpans( int numsamples, CSentence& outwords )
+{
+	int numwords = outwords.m_Words.Size();
+	// Nothing to do
+	if ( numwords <= 0 )
+		return;
+
+	int interationcount = 1;
+
+	while( 1 )
+	{
+		Log( "\nCompute %i\n", interationcount++ );
+		LogWords( outwords );
+
+		int wordNumber;
+
+		// Done!
+		if ( !CountUnuseableWords( outwords ) )
+		{
+			FixupZeroLengthWords( outwords );
+			break;
+		}
+
+		if ( !CountUsableWords( outwords ) )
+		{
+			// Evenly space words across full sample time
+			PartitionWords( outwords, 0, numwords - 1, 0, numsamples );
+			break;
+		}
+
+		wordNumber = FindFirstUsableWord( outwords );
+		// Not the first word
+		if ( wordNumber > 0 )
+		{
+			// Repartition all of the unusables and the first one starting at zero over the range
+			CWordTag *firstUsable = outwords.m_Words[ wordNumber ];
+			Assert( firstUsable );
+
+			if ( firstUsable->m_uiStartByte != 0 )
+			{
+				PartitionWords( outwords, 0, wordNumber - 1, 0, firstUsable->m_uiStartByte );
+			}
+			else
+			{
+				PartitionWords( outwords, 0, wordNumber, 0, firstUsable->m_uiEndByte );
+			}
+
+			// Start over
+			continue;
+		}
+
+		wordNumber = FindLastUsableWord( outwords );
+		// Not the last word
+		if ( wordNumber >= 0 && wordNumber < numwords - 1 )
+		{
+			// Repartition all of the unusables and the first one starting at zero over the range
+			CWordTag *lastUsable = outwords.m_Words[ wordNumber ];
+			Assert( lastUsable );
+
+			if ( lastUsable->m_uiEndByte != (unsigned int)numsamples )
+			{
+				PartitionWords( outwords, wordNumber + 1, numwords-1, lastUsable->m_uiEndByte, numsamples );
+			}
+			else
+			{
+				PartitionWords( outwords, wordNumber, numwords-1, lastUsable->m_uiStartByte, numsamples );
+			}
+
+			// Start over
+			continue;
+		}
+
+		// If we get here it means that the start and end of the list are okay and we just have to
+		//  iterate across the list and fix things in the middle
+		int startByte = 0;
+		int endByte = 0;
+		for ( int i = 0; i < numwords ; i++ )
+		{
+			CWordTag *word = outwords.m_Words[ i ];
+			if ( IsUseable( word ) )
+			{
+				startByte = word->m_uiEndByte;
+				continue;
+			}
+
+			// Found the start of a chain of 1 or more unusable words
+			// Find the startbyte of the next usable word and count how many words we check
+			int wordCount = 1;
+			for ( int j = i + 1; j < numwords; j++ )
+			{
+				CWordTag *next = outwords.m_Words[ j ];
+				if ( IsUseable( next ) )
+				{
+					endByte = next->m_uiStartByte;
+					break;
+				}
+
+				wordCount++;
+			}
+
+			// Now partition words across the gap and go to start again
+			PartitionWords( outwords, i, i + wordCount - 1, startByte, endByte );
+			break;
+		}
+	}
+}
+
+//-----------------------------------------------------------------------------
+// Purpose: Given a wavfile and a list of inwords, determines the word/phonene 
+//  sample counts for the sentce
+// Input  : *wavfile - 
+//			*inwords - 
+//			*outphonemes{	text.Clear( - 
+// Output : SR_RESULT
+//-----------------------------------------------------------------------------
+static SR_RESULT SAPI_ExtractPhonemes( 
+	const char *wavfile,
+	int numsamples,
+	void (*pfnPrint)( const char *fmt, ... ),
+	CSentence& inwords,
+	CSentence& outwords )
+{
+	LogReset();
+
+	USES_CONVERSION;
+
+	CSpDynamicString text;
+	text.Clear();
+
+	HKEY hkwipe;
+	LONG lResult = RegOpenKeyEx( HKEY_CURRENT_USER, "Software\\Microsoft\\Speech\\RecoProfiles", 0, KEY_ALL_ACCESS, &hkwipe );
+	if ( lResult == ERROR_SUCCESS )
+	{
+		RecursiveRegDelKey( hkwipe );
+		RegCloseKey( hkwipe );
+	}
+
+	if ( strlen( inwords.GetText() ) <= 0 )
+	{
+		inwords.SetTextFromWords();
+	}
+
+	// Construct a string from the inwords array
+	text.Append( T2W( inwords.GetText() ) );
+
+	// Assume failure
+	SR_RESULT result = SR_RESULT_ERROR;
+
+	if ( text.Length() > 0 )
+	{
+		CSentence sentence;
+
+		pfnPrint( "Processing...\r\n" );
+
+		// Give it a try
+		result = ExtractPhonemes( wavfile, text, sentence, pfnPrint );
+
+		pfnPrint( "Finished.\r\n" );
+		// PrintWordsAndPhonemes( sentence, pfnPrint );
+
+		// Copy results to outputs
+		outwords.Reset();
+
+		outwords.SetText( inwords.GetText() );
+		
+		Log( "Starting\n" );
+		LogWords( inwords );
+
+		if ( SR_RESULT_ERROR != result )
+		{
+			int i;
+
+			Log( "Hypothesized\n" );
+			LogWords( sentence );
+
+			for( i = 0 ; i < sentence.m_Words.Size(); i++ )
+			{
+				CWordTag *tag = sentence.m_Words[ i ];
+				if ( tag )
+				{
+					// Skip '...' tag
+					if ( stricmp( tag->GetWord(), "..." ) )
+					{
+						CWordTag *newTag = new CWordTag( *tag );
+
+						outwords.m_Words.AddToTail( newTag );
+					}
+				}
+			}
+
+			// Now insert unrecognized/skipped words from original list
+			//
+			int frompos = 0, topos = 0;
+
+			while( 1 )
+			{
+				// End of source list
+				if ( frompos >= inwords.m_Words.Size() )
+					break;
+
+				const CWordTag *fromTag = inwords.m_Words[ frompos ];
+
+				// Reached end of destination list, just copy words over from from source list until
+				//  we run out of source words
+				if ( topos >= outwords.m_Words.Size() )
+				{
+					// Just copy words over
+					CWordTag *newWord = new CWordTag( *fromTag );
+
+					// Remove phonemes
+					while ( newWord->m_Phonemes.Size() > 0 )
+					{
+						CPhonemeTag *kill = newWord->m_Phonemes[ 0 ];
+						newWord->m_Phonemes.Remove( 0 );
+						delete kill;
+					}
+
+					outwords.m_Words.AddToTail( newWord );
+					frompos++;
+					topos++;
+					continue;
+				}
+
+				// Destination word
+				const CWordTag *toTag = outwords.m_Words[ topos ];
+
+				// Words match, just skip ahead
+				if ( !stricmp( fromTag->GetWord(), toTag->GetWord() ) )
+				{
+					frompos++;
+					topos++;
+					continue;
+				}
+
+				// The only case we handle is that something in the source wasn't in the destination
+
+				// Find the next source word that appears in the destination
+				int skipAhead = frompos + 1;
+				bool found = false;
+				while ( skipAhead < inwords.m_Words.Size() )
+				{
+					const CWordTag *sourceWord = inwords.m_Words[ skipAhead ];
+					if ( !stricmp( sourceWord->GetWord(), toTag->GetWord() ) )
+					{
+						found = true;
+						break;
+					}
+
+					skipAhead++;
+				}
+
+				// Uh oh destination has words that are not in source, just skip to next destination word?
+				if ( !found )
+				{
+					topos++;
+				}
+				else
+				{
+					// Copy words from from source list into destination
+					// 
+					int skipCount = skipAhead - frompos;
+
+					while ( --skipCount>= 0 )
+					{
+						const CWordTag *sourceWord = inwords.m_Words[ frompos++ ];
+						CWordTag *newWord = new CWordTag( *sourceWord );
+
+						// Remove phonemes
+						while ( newWord->m_Phonemes.Size() > 0 )
+						{
+							CPhonemeTag *kill = newWord->m_Phonemes[ 0 ];
+							newWord->m_Phonemes.Remove( 0 );
+							delete kill;
+						}
+
+						outwords.m_Words.InsertBefore( topos, newWord );
+						topos++;
+					}
+
+					frompos++;
+					topos++;
+				}
+			}
+
+			Log( "\nDone simple check\n" );
+
+			LogWords( outwords );
+			LogPhonemes( outwords );
+
+			ComputeMissingByteSpans( numsamples, outwords );
+
+			Log( "\nFinal check\n" );
+
+			LogWords( outwords );
+			LogPhonemes( outwords );
+		}
+	}
+	else
+	{
+		pfnPrint( "Input sentence is empty!\n" );
+	}
+
+	// Return results
+	return result;
+}
+
+
+//-----------------------------------------------------------------------------
+// Purpose: Expose the interface
+//-----------------------------------------------------------------------------
+class CPhonemeExtractorSAPI : public IPhonemeExtractor
+{
+public:
+	virtual PE_APITYPE	GetAPIType() const
+	{
+		return SPEECH_API_SAPI;
+	}
+
+	// Used for menus, etc
+	virtual char const *GetName() const
+	{
+		return "MS SAPI 5.1";
+	}
+
+	SR_RESULT Extract( 
+		const char *wavfile,
+		int numsamples,
+		void (*pfnPrint)( const char *fmt, ... ),
+		CSentence& inwords,
+		CSentence& outwords )
+	{
+		return SAPI_ExtractPhonemes( wavfile, numsamples, pfnPrint, inwords, outwords );
+	}
+};
+
 EXPOSE_SINGLE_INTERFACE( CPhonemeExtractorSAPI, IPhonemeExtractor, VPHONEME_EXTRACTOR_INTERFACE );
\ No newline at end of file
diff --git a/mp/src/utils/phonemeextractor/phonemeextractor.vpc b/mp/src/utils/phonemeextractor/phonemeextractor.vpc
index f6e82e4d..cdcee5c2 100644
--- a/mp/src/utils/phonemeextractor/phonemeextractor.vpc
+++ b/mp/src/utils/phonemeextractor/phonemeextractor.vpc
@@ -1,83 +1,83 @@
-//-----------------------------------------------------------------------------

-// PHONEMEEXTRACTOR.VPC

-//

-//	Project Script

-//-----------------------------------------------------------------------------

-

-$Macro SRCDIR		"..\.."

-$Macro OUTBINDIR	"$SRCDIR\..\game\bin\phonemeextractors"

-

-$Include "$SRCDIR\vpc_scripts\source_dll_base.vpc"

-

-$Configuration

-{

-	$Compiler

-	{

-		$AdditionalIncludeDirectories		"$BASE;../common,../hlfaceposer,../sapi51/include"

-		$PreprocessorDefinitions			"$BASE;PHONEMEEXTRACTOR_EXPORTS"

-

-		// The project has some trouble with the deprecated string function warnings, so turn those off.

-		$AdditionalOptions	"$BASE /wd4995"

-	}

-	

-	$Linker

-	{

-		$AdditionalDependencies				"$BASE odbc32.lib odbccp32.lib"

-	}

-}

-

-$Project "Phonemeextractor"

-{

-	$Folder	"Source Files"

-	{

-		$File	"extractor_utils.cpp"

-		$File	"$SRCDIR\public\phonemeconverter.cpp"

-		$File	"$SRCDIR\public\sentence.cpp"

-		$File	"phonemeextractor.cpp"

-	}

-

-	$Folder	"Header Files"

-	{

-		$File	"talkback.h"

-	}

-	

-	$Folder	"SAPI Header Files"

-	{ // These are dynamic because sapi might not be present in SDK branches, but we still want VPC to succeed.

-		$DynamicFile	"..\sapi51\Include\sapi.h"

-		$DynamicFile	"..\sapi51\Include\sapiddk.h"

-		$DynamicFile	"..\sapi51\Include\Spddkhlp.h"

-		$DynamicFile	"..\sapi51\Include\spdebug.h"

-		$DynamicFile	"..\sapi51\Include\sperror.h"

-		$DynamicFile	"..\sapi51\Include\sphelper.h"

-	}

-	

-	$Folder	"Public Header Files"

-	{

-		$File	"$SRCDIR\public\mathlib\amd3dx.h"

-		$File	"$SRCDIR\public\tier0\basetypes.h"

-		$File	"$SRCDIR\public\tier0\commonmacros.h"

-		$File	"$SRCDIR\public\tier0\dbg.h"

-		$File	"$SRCDIR\public\tier0\fasttimer.h"

-		$File	"$SRCDIR\public\appframework\IAppSystem.h"

-		$File	"$SRCDIR\public\mathlib\mathlib.h"

-		$File	"$SRCDIR\public\phonemeconverter.h"

-		$File	"$SRCDIR\public\phonemeextractor\phonemeextractor.h"

-		$File	"$SRCDIR\public\tier0\platform.h"

-		$File	"$SRCDIR\public\tier0\protected_things.h"

-		$File	"$SRCDIR\public\sentence.h"

-		$File	"$SRCDIR\public\string_t.h"

-		$File	"$SRCDIR\public\tier1\strtools.h"

-		$File	"$SRCDIR\public\tier1\utllinkedlist.h"

-		$File	"$SRCDIR\public\tier1\utlmemory.h"

-		$File	"$SRCDIR\public\tier1\utlvector.h"

-		$File	"$SRCDIR\public\mathlib\vector.h"

-		$File	"$SRCDIR\public\mathlib\vector2d.h"

-		$File	"$SRCDIR\public\vstdlib\vstdlib.h"

-	}

-	

-	$Folder	"Link Libraries"

-	{

-		$Lib mathlib

-		$DynamicFile	"..\sapi51\lib\i386\sapi.lib"

-	}

-}

+//-----------------------------------------------------------------------------
+// PHONEMEEXTRACTOR.VPC
+//
+//	Project Script
+//-----------------------------------------------------------------------------
+
+$Macro SRCDIR		"..\.."
+$Macro OUTBINDIR	"$SRCDIR\..\game\bin\phonemeextractors"
+
+$Include "$SRCDIR\vpc_scripts\source_dll_base.vpc"
+
+$Configuration
+{
+	$Compiler
+	{
+		$AdditionalIncludeDirectories		"$BASE;../common,../hlfaceposer,../sapi51/include"
+		$PreprocessorDefinitions			"$BASE;PHONEMEEXTRACTOR_EXPORTS"
+
+		// The project has some trouble with the deprecated string function warnings, so turn those off.
+		$AdditionalOptions	"$BASE /wd4995"
+	}
+	
+	$Linker
+	{
+		$AdditionalDependencies				"$BASE odbc32.lib odbccp32.lib"
+	}
+}
+
+$Project "Phonemeextractor"
+{
+	$Folder	"Source Files"
+	{
+		$File	"extractor_utils.cpp"
+		$File	"$SRCDIR\public\phonemeconverter.cpp"
+		$File	"$SRCDIR\public\sentence.cpp"
+		$File	"phonemeextractor.cpp"
+	}
+
+	$Folder	"Header Files"
+	{
+		$File	"talkback.h"
+	}
+	
+	$Folder	"SAPI Header Files"
+	{ // These are dynamic because sapi might not be present in SDK branches, but we still want VPC to succeed.
+		$DynamicFile	"..\sapi51\Include\sapi.h"
+		$DynamicFile	"..\sapi51\Include\sapiddk.h"
+		$DynamicFile	"..\sapi51\Include\Spddkhlp.h"
+		$DynamicFile	"..\sapi51\Include\spdebug.h"
+		$DynamicFile	"..\sapi51\Include\sperror.h"
+		$DynamicFile	"..\sapi51\Include\sphelper.h"
+	}
+	
+	$Folder	"Public Header Files"
+	{
+		$File	"$SRCDIR\public\mathlib\amd3dx.h"
+		$File	"$SRCDIR\public\tier0\basetypes.h"
+		$File	"$SRCDIR\public\tier0\commonmacros.h"
+		$File	"$SRCDIR\public\tier0\dbg.h"
+		$File	"$SRCDIR\public\tier0\fasttimer.h"
+		$File	"$SRCDIR\public\appframework\IAppSystem.h"
+		$File	"$SRCDIR\public\mathlib\mathlib.h"
+		$File	"$SRCDIR\public\phonemeconverter.h"
+		$File	"$SRCDIR\public\phonemeextractor\phonemeextractor.h"
+		$File	"$SRCDIR\public\tier0\platform.h"
+		$File	"$SRCDIR\public\tier0\protected_things.h"
+		$File	"$SRCDIR\public\sentence.h"
+		$File	"$SRCDIR\public\string_t.h"
+		$File	"$SRCDIR\public\tier1\strtools.h"
+		$File	"$SRCDIR\public\tier1\utllinkedlist.h"
+		$File	"$SRCDIR\public\tier1\utlmemory.h"
+		$File	"$SRCDIR\public\tier1\utlvector.h"
+		$File	"$SRCDIR\public\mathlib\vector.h"
+		$File	"$SRCDIR\public\mathlib\vector2d.h"
+		$File	"$SRCDIR\public\vstdlib\vstdlib.h"
+	}
+	
+	$Folder	"Link Libraries"
+	{
+		$Lib mathlib
+		$DynamicFile	"..\sapi51\lib\i386\sapi.lib"
+	}
+}
diff --git a/mp/src/utils/phonemeextractor/phonemeextractor_ims.cpp b/mp/src/utils/phonemeextractor/phonemeextractor_ims.cpp
index 70819f8e..29dabab4 100644
--- a/mp/src/utils/phonemeextractor/phonemeextractor_ims.cpp
+++ b/mp/src/utils/phonemeextractor/phonemeextractor_ims.cpp
@@ -1,1075 +1,1075 @@
-//========= Copyright Valve Corporation, All rights reserved. ============//

-//

-// Purpose: 

-//

-// $NoKeywords: $

-//=============================================================================//

-#include <stdio.h>

-#include <stdarg.h>

-#include <memory.h>

-#include <windows.h>

-#include <mmsystem.h>

-#include <mmreg.h>

-#include <sys/types.h>

-#include <sys/stat.h>

-

-#include "phonemeextractor/PhonemeExtractor.h"

-#include "ims_helper/ims_helper.h"

-

-#include "tier0/dbg.h"

-#include "sentence.h"

-#include "PhonemeConverter.h"

-#include "tier1/strtools.h"

-

-#define TEXTLESS_WORDNAME	"[Textless]"

-

-static IImsHelper *talkback = NULL;

-

-//-----------------------------------------------------------------------------

-// Purpose: Expose the interface

-//-----------------------------------------------------------------------------

-class CPhonemeExtractorLipSinc : public IPhonemeExtractor

-{

-public:

-	virtual PE_APITYPE	GetAPIType() const

-	{

-		return SPEECH_API_LIPSINC;

-	}

-

-	// Used for menus, etc

-	virtual char const *GetName() const

-	{

-		return "IMS (LipSinc)";

-	}

-

-	SR_RESULT Extract( 

-		const char *wavfile,

-		int numsamples,

-		void (*pfnPrint)( const char *fmt, ... ),

-		CSentence& inwords,

-		CSentence& outwords );

-

-

-	CPhonemeExtractorLipSinc( void );

-	~CPhonemeExtractorLipSinc( void );

-

-	enum

-	{

-		MAX_WORD_LENGTH = 128,

-	};

-private:

-

-

-	class CAnalyzedWord

-	{

-	public:

-		char		buffer[ MAX_WORD_LENGTH ];

-		double		starttime;

-		double		endtime;

-	};

-

-	class CAnalyzedPhoneme

-	{

-	public:

-		char		phoneme[ 32 ];

-		double		starttime;

-		double		endtime;

-	};

-

-	bool InitLipSinc( void );

-	void ShutdownLipSinc( void );

-

-	void DescribeError( TALKBACK_ERR err );

-	void Printf( char const *fmt, ... );

-

-	bool CheckSoundFile( char const *filename );

-	bool GetInitialized( void );

-	void SetInitialized( bool init );

-

-	void (*m_pfnPrint)( const char *fmt, ... );

-

-	char const *ConstructInputSentence( CSentence& inwords );

-	bool AttemptAnalysis( TALKBACK_ANALYSIS **ppAnalysis, char const *wavfile, CSentence& inwords );

-

-	char const *ApplyTBWordRules( char const *word );

-

-	void ProcessWords( TALKBACK_ANALYSIS *analysis, CSentence& inwords, CSentence& outwords );

-	void ProcessWordsTextless( TALKBACK_ANALYSIS *analysis, CSentence& outwords );

-

-	int GetPhonemeIndexAtWord( TALKBACK_ANALYSIS *analysis, double time, bool checkstart );

-

-	int GetPhonemeIndexAtWordStart( TALKBACK_ANALYSIS *analysis, double starttime );

-	int GetPhonemeIndexAtWordEnd( TALKBACK_ANALYSIS *analysis, double endtime );

-

-	CAnalyzedWord *GetAnalyzedWord( TALKBACK_ANALYSIS *analysis, int index );

-	CAnalyzedPhoneme *GetAnalyzedPhoneme( TALKBACK_ANALYSIS *analysis, int index );

-

-	int ComputeByteFromTime( float time );

-

-	bool m_bInitialized;

-

-	float	m_flSampleCount;

-	float	m_flDuration;

-

-	float	m_flSamplesPerSecond;

-

-	int		m_nBytesPerSample;

-

-	HMODULE m_hHelper;

-};

-

-CPhonemeExtractorLipSinc::CPhonemeExtractorLipSinc( void )

-{

-	m_hHelper = (HMODULE)0;

-	m_pfnPrint = NULL;

-

-	m_bInitialized = false;

-	

-	m_flSampleCount = 0.0f;

-	m_flDuration = 0.0f;

-

-	m_flSamplesPerSecond = 0.0f;

-

-	m_nBytesPerSample = 0;

-}

-

-CPhonemeExtractorLipSinc::~CPhonemeExtractorLipSinc( void )

-{

-	if ( GetInitialized() )

-	{

-		ShutdownLipSinc();

-	}

-}

-

-bool CPhonemeExtractorLipSinc::GetInitialized( void )

-{

-	return m_bInitialized;

-}

-

-void CPhonemeExtractorLipSinc::SetInitialized( bool init )

-{

-	m_bInitialized = init;

-}

-

-int CPhonemeExtractorLipSinc::ComputeByteFromTime( float time )

-{

-	if ( !m_flDuration )

-		return 0;

-

-	float frac = time / m_flDuration;

-

-	float sampleNumber = frac * m_flSampleCount;

-

-	int bytenumber = sampleNumber * m_nBytesPerSample;

-

-	return bytenumber;

-}

-

-void CPhonemeExtractorLipSinc::DescribeError( TALKBACK_ERR err )

-{

-	Assert( m_pfnPrint );

-

-	// Get the error description.

-	char errorDesc[256] = "";

-	if ( err != TALKBACK_NOERR )

-	{

-		talkback->TalkBackGetErrorString( err, sizeof(errorDesc), errorDesc );

-	}

-	

-	// Report or log the error...

-	(*m_pfnPrint)( "LIPSINC ERROR:  %s\n", errorDesc );

-}

-

-//-----------------------------------------------------------------------------

-// Purpose: 

-// Input  : *fmt - 

-//			.. - 

-//-----------------------------------------------------------------------------

-void CPhonemeExtractorLipSinc::Printf( char const *fmt, ... )

-{

-	Assert( m_pfnPrint );

-

-	char string[ 4096 ];

-

-	va_list argptr;

-	va_start( argptr, fmt );

-	vsprintf( string, fmt, argptr );

-	va_end( argptr );

-

-	(*m_pfnPrint)( "%s", string );

-}

-

-bool CPhonemeExtractorLipSinc::CheckSoundFile( char const *filename )

-{

-	TALKBACK_SOUND_FILE_METRICS fm;

-	memset( &fm, 0, sizeof( fm ) );

-	fm.m_size = sizeof( fm );

-

-	TALKBACK_ERR err = talkback->TalkBackGetSoundFileMetrics( filename, &fm );

-	if ( err != TALKBACK_NOERR )

-	{

-		DescribeError( err );

-		return false;

-	}

-

-	if ( fm.m_canBeAnalyzed )

-	{

-		Printf( "%s:  %.2f s, rate %i, bits %i, channels %i\n",

-			filename,

-			fm.m_duration,

-			fm.m_sampleRate,

-			fm.m_bitsPerSample,

-			fm.m_channelCount );

-	}

-

-	m_flDuration = fm.m_duration;

-	if ( m_flDuration > 0 )

-	{

-		m_flSamplesPerSecond = m_flSampleCount / m_flDuration;

-	}

-	else

-	{

-		m_flSamplesPerSecond = 0.0f;

-	}

-

-	m_nBytesPerSample = ( fm.m_bitsPerSample >> 3 );

-

-	m_flSampleCount /= m_nBytesPerSample;

-

-	m_nBytesPerSample /= fm.m_channelCount;

-

-	return fm.m_canBeAnalyzed ? true : false;

-}

-

-typedef IImsHelper *(*pfnImsHelper)(void);

-

-//-----------------------------------------------------------------------------

-// Purpose: 

-// Output : Returns true on success, false on failure.

-//-----------------------------------------------------------------------------

-bool CPhonemeExtractorLipSinc::InitLipSinc( void )

-{

-	if ( GetInitialized() )

-	{

-		return true;

-	}

-

-	m_hHelper = LoadLibrary( "ims_helper.dll" );

-	if ( !m_hHelper )

-	{

-		return false;

-	}

-	

-	pfnImsHelper factory = (pfnImsHelper)::GetProcAddress( m_hHelper, "GetImsHelper" );

-	if ( !factory )

-	{

-		FreeLibrary( m_hHelper );

-		return false;

-	}

-

-	talkback = reinterpret_cast< IImsHelper * >( (*factory)() );

-	if ( !talkback )

-	{

-		FreeLibrary( m_hHelper );

-		return false;

-	}

-

-	char szExeName[ MAX_PATH ];

-	szExeName[0] = 0;

-	GetModuleFileName( (HMODULE)0, szExeName, sizeof( szExeName ) );

-

-	char szBaseDir[ MAX_PATH ];

-	Q_strncpy( szBaseDir, szExeName, sizeof( szBaseDir ) );

-

-	Q_StripLastDir( szBaseDir, sizeof( szBaseDir ) );

-	Q_StripTrailingSlash( szBaseDir );

-	Q_strlower( szBaseDir );

-

-	char coreDataDir[ 512 ];

-	Q_snprintf( coreDataDir, sizeof( coreDataDir ), "%s\\lipsinc_data\\",

-		szBaseDir );

-	Q_FixSlashes( coreDataDir );

-

-	char szCheck[ 512 ];

-	Q_snprintf( szCheck, sizeof( szCheck ), "%sDtC6dal.dat", coreDataDir );

-	struct __stat64 buf;

-

-	if ( _stat64( szCheck, &buf ) != 0 )

-	{

-		Q_snprintf( coreDataDir, sizeof( coreDataDir ), "%s\\bin\\lipsinc_data\\",

-			szBaseDir );

-		Q_FixSlashes( coreDataDir );

-		Q_snprintf( szCheck, sizeof( szCheck ), "%sDtC6dal.dat", coreDataDir );

-

-		if ( _stat64( szCheck, &buf ) != 0 )

-		{

-			Error( "Unable to find talkback data files in %s.", coreDataDir );

-		}

-	}

-

-	TALKBACK_ERR err;

-	

-	err = talkback->TalkBackStartupLibrary( coreDataDir );

-	if ( err != TALKBACK_NOERR )

-	{

-		DescribeError( err );

-		FreeLibrary( m_hHelper );

-		return false;

-	}

-

-	long verMajor = 0;

-	long verMinor = 0;

-	long verRevision = 0;

-	

-	err = talkback->TalkBackGetVersion(

-		&verMajor, 

-		&verMinor, 

-		&verRevision);

-	if ( err != TALKBACK_NOERR )

-	{

-		DescribeError( err );

-		FreeLibrary( m_hHelper );

-		return false;

-	}

-

-	Printf( "Lipsinc TalkBack Version %i.%i.%i\n", verMajor, verMinor, verRevision );

-

-	m_bInitialized = true;

-

-	return true;

-}

-

-//-----------------------------------------------------------------------------

-// Purpose: 

-//-----------------------------------------------------------------------------

-void CPhonemeExtractorLipSinc::ShutdownLipSinc( void )

-{

-	// HACK HACK:  This seems to crash on exit sometimes

-	__try

-	{

-		talkback->TalkBackShutdownLibrary();

-

-		FreeLibrary( m_hHelper );

-	}

-	__except(EXCEPTION_EXECUTE_HANDLER )

-	{

-		OutputDebugString( "----> Crash shutting down TALKBACK sdk, exception caught and ignored\n" );

-	}

-}

-

-//-----------------------------------------------------------------------------

-// Purpose: 

-// Input  : inwords - 

-// Output : char const

-//-----------------------------------------------------------------------------

-char const *CPhonemeExtractorLipSinc::ConstructInputSentence( CSentence& inwords )

-{

-	static char sentence[ 16384 ];

-

-	sentence[ 0 ] = 0;

-

-	int last = inwords.m_Words.Size() - 1;

-

-	for ( int i = 0 ; i <= last; i++ )

-	{

-		CWordTag *w = inwords.m_Words[ i ];

-

-		strcat( sentence, w->GetWord() );

-		if ( i != last )

-		{

-			strcat( sentence, " " );

-		}

-	}

-

-	if ( inwords.m_Words.Count() == 1 && 

-		!Q_strnicmp( inwords.GetText(), TEXTLESS_WORDNAME, Q_strlen( TEXTLESS_WORDNAME ) ) )

-	{

-		sentence[ 0 ] = 0;

-	}

-

-	return sentence;

-}

-

-bool CPhonemeExtractorLipSinc::AttemptAnalysis( TALKBACK_ANALYSIS **ppAnalysis, char const *wavfile, CSentence& inwords )

-{

-	*ppAnalysis = NULL;

-

-	TALKBACK_ANALYSIS_SETTINGS settings;

-	memset( &settings, 0, sizeof( settings ) );

-

-   // Set this field to sizeof(TALKBACK_ANALYSIS_SETTINGS) before using the

-    // structure.

-	settings.fSize = sizeof( TALKBACK_ANALYSIS_SETTINGS );

-	

-

-    // Default value: 30 (frames per second).

-    settings.fFrameRate = 100;

-    // Set this to 1 to optimize for flipbook output, 0 to do analysis normally.

-    //

-    // Default value: 0 (normal analysis).

-    settings.fOptimizeForFlipbook = 0;

-    // Set this to -1 to seed the random number generator with the current time.

-    // Any other number will be used directly for the random number seed, which

-    // is useful if you want repeatable speech gestures. This value does not

-    // influence lip-synching at all.

-    //

-    // Default value: -1 (use current time).

-    settings.fRandomSeed = -1;

-    // Path to the configuration (.INI) file with phoneme-to-speech-target

-    // mapping. Set this to NULL to use the default mapping.

-    //

-    // Default value: NULL (use default mapping).

-    settings.fConfigFile = NULL;

-

-	char const *text = ConstructInputSentence( inwords );

-

-	Printf( "Analyzing: \"%s\"\n", text[ 0 ] ? text : TEXTLESS_WORDNAME );

-

-	TALKBACK_ERR err = talkback->TalkBackGetAnalysis( 

-		ppAnalysis,

-		wavfile,

-		text,

-		&settings );

-

-	if ( err != TALKBACK_NOERR )

-	{

-		DescribeError( err );

-		return false;

-	}

-

-	Printf( "Analysis successful...\n" );

-

-	return true;

-}

-

-typedef struct

-{

-	TALKBACK_PHONEME phoneme;

-	char const		*string;

-} TBPHONEMES_t;

-

-static TBPHONEMES_t g_TBPhonemeList[]=

-{

-	{ TALKBACK_PHONEME_IY, "iy" },

-	{ TALKBACK_PHONEME_IH, "ih" },

-	{ TALKBACK_PHONEME_EH, "eh" },

-	{ TALKBACK_PHONEME_EY, "ey" },

-	{ TALKBACK_PHONEME_AE, "ae" },

-	{ TALKBACK_PHONEME_AA, "aa" },

-	{ TALKBACK_PHONEME_AW, "aw" },

-	{ TALKBACK_PHONEME_AY, "ay" },

-	{ TALKBACK_PHONEME_AH, "ah" },

-	{ TALKBACK_PHONEME_AO, "ao" },

-	{ TALKBACK_PHONEME_OY, "oy" },

-	{ TALKBACK_PHONEME_OW, "ow" },

-	{ TALKBACK_PHONEME_UH, "uh" },

-	{ TALKBACK_PHONEME_UW, "uw" },

-	{ TALKBACK_PHONEME_ER, "er" },

-	{ TALKBACK_PHONEME_AX, "ax" },

-	{ TALKBACK_PHONEME_S, "s" },

-	{ TALKBACK_PHONEME_SH, "sh" },

-	{ TALKBACK_PHONEME_Z, "z" },

-	{ TALKBACK_PHONEME_ZH, "zh" },

-	{ TALKBACK_PHONEME_F, "f" },

-	{ TALKBACK_PHONEME_TH, "th" },

-	{ TALKBACK_PHONEME_V, "v" },

-	{ TALKBACK_PHONEME_DH, "dh" },

-	{ TALKBACK_PHONEME_M, "m" },

-	{ TALKBACK_PHONEME_N, "n" },

-	{ TALKBACK_PHONEME_NG, "ng" },

-	{ TALKBACK_PHONEME_L, "l" },

-	{ TALKBACK_PHONEME_R, "r" },

-	{ TALKBACK_PHONEME_W, "w" },

-	{ TALKBACK_PHONEME_Y, "y" },

-	{ TALKBACK_PHONEME_HH, "hh" },

-	{ TALKBACK_PHONEME_B, "b" },

-	{ TALKBACK_PHONEME_D, "d" },

-	{ TALKBACK_PHONEME_JH, "jh" },

-	{ TALKBACK_PHONEME_G, "g" },

-	{ TALKBACK_PHONEME_P, "p" },

-	{ TALKBACK_PHONEME_T, "t" },

-	{ TALKBACK_PHONEME_K, "k" },

-	{ TALKBACK_PHONEME_CH, "ch" },

-	{ TALKBACK_PHONEME_SIL, "<sil>" },

-	{ -1, NULL }

-};

-

-char const *TBPhonemeToString( TALKBACK_PHONEME phoneme )

-{

-	if ( phoneme < TALKBACK_PHONEME_FIRST || phoneme > TALKBACK_PHONEME_LAST )

-	{

-		return "Bogus";

-	}

-

-	TBPHONEMES_t *item = &g_TBPhonemeList[ phoneme ];

-	return item->string;

-}

-

-//-----------------------------------------------------------------------------

-// Purpose: 

-// Input  : *analysis - 

-//			time - 

-//			start - 

-// Output : int

-//-----------------------------------------------------------------------------

-int CPhonemeExtractorLipSinc::GetPhonemeIndexAtWord( TALKBACK_ANALYSIS *analysis, double time, bool start )

-{

-	long count;

-

-	TALKBACK_ERR err = talkback->TalkBackGetNumPhonemes( analysis, &count );

-	if ( err != TALKBACK_NOERR )

-	{

-		DescribeError( err );

-		return -1;

-	}

-

-	if ( count <= 0L )

-		return -1;

-

-	// Bogus

-	if ( count >= 100000L )

-		return -1;

-

-	for ( int i = 0; i < (int)count; i++ )

-	{

-		TALKBACK_PHONEME tbPhoneme = TALKBACK_PHONEME_INVALID;

-		err = talkback->TalkBackGetPhonemeEnum( analysis, i, &tbPhoneme );

-		if ( err != TALKBACK_NOERR )

-		{

-			DescribeError( err );

-			continue;

-		}

-

-		double t;

-

-		if ( start )

-		{

-			err = talkback->TalkBackGetPhonemeStartTime( analysis, i, &t );

-		}

-		else

-		{

-			err = talkback->TalkBackGetPhonemeEndTime( analysis, i, &t );

-		}

-

-		if ( err != TALKBACK_NOERR )

-		{

-			DescribeError( err );

-			continue;

-		}

-

-		if ( t == time )

-		{

-			return i;

-		}

-	}

-

-	return -1;

-}

-

-//-----------------------------------------------------------------------------

-// Purpose: 

-// Input  : *analysis - 

-//			starttime - 

-// Output : int

-//-----------------------------------------------------------------------------

-int CPhonemeExtractorLipSinc::GetPhonemeIndexAtWordStart( TALKBACK_ANALYSIS *analysis, double starttime )

-{

-	return GetPhonemeIndexAtWord( analysis, starttime, true );

-}

-

-//-----------------------------------------------------------------------------

-// Purpose: 

-// Input  : *analysis - 

-//			endtime - 

-// Output : int

-//-----------------------------------------------------------------------------

-int CPhonemeExtractorLipSinc::GetPhonemeIndexAtWordEnd( TALKBACK_ANALYSIS *analysis, double endtime )

-{

-	return GetPhonemeIndexAtWord( analysis, endtime, false );

-}

-

-CPhonemeExtractorLipSinc::CAnalyzedPhoneme *CPhonemeExtractorLipSinc::GetAnalyzedPhoneme( TALKBACK_ANALYSIS *analysis, int index )

-{

-	static CAnalyzedPhoneme p;

-

-	memset( &p, 0, sizeof( p ) );

-

-	TALKBACK_PHONEME tb;

-

-	TALKBACK_ERR err = talkback->TalkBackGetPhonemeEnum( analysis, index, &tb );

-	if ( err != TALKBACK_NOERR )

-	{

-		DescribeError( err );

-		return NULL;

-	}

-

-	strcpy( p.phoneme, TBPhonemeToString( tb ) );

-

-	err = talkback->TalkBackGetPhonemeStartTime( analysis, index, &p.starttime );

-	if ( err != TALKBACK_NOERR )

-	{

-		DescribeError( err );

-		return NULL;

-	}

-	err = talkback->TalkBackGetPhonemeEndTime( analysis, index, &p.endtime );

-	if ( err != TALKBACK_NOERR )

-	{

-		DescribeError( err );

-		return NULL;

-	}

-

-	return &p;

-}

-

-CPhonemeExtractorLipSinc::CAnalyzedWord *CPhonemeExtractorLipSinc::GetAnalyzedWord( TALKBACK_ANALYSIS *analysis, int index )

-{

-	static CAnalyzedWord w;

-

-	memset( &w, 0, sizeof( w ) );

-

-	long chars = sizeof( w.buffer );

-

-	TALKBACK_ERR err = talkback->TalkBackGetWord( analysis, index, chars, w.buffer );

-	if ( err != TALKBACK_NOERR )

-	{

-		DescribeError( err );

-		return NULL;

-	}

-

-	err = talkback->TalkBackGetWordStartTime( analysis, index, &w.starttime );

-	if ( err != TALKBACK_NOERR )

-	{

-		DescribeError( err );

-		return NULL;

-	}

-	err = talkback->TalkBackGetWordEndTime( analysis, index, &w.endtime );

-	if ( err != TALKBACK_NOERR )

-	{

-		DescribeError( err );

-		return NULL;

-	}

-

-	return &w;

-}

-

-//-----------------------------------------------------------------------------

-// Purpose: 

-// Input  : *w1 - 

-//			*w2 - 

-// Output : Returns true on success, false on failure.

-//-----------------------------------------------------------------------------

-bool FuzzyWordMatch( char const *w1, char const *w2 )

-{

-	int len1 = strlen( w1 );

-	int len2 = strlen( w2 );

-

-	int minlen = min( len1, len2 );

-

-	// Found a match

-	if ( !strnicmp( w1, w2, minlen ) )

-		return true;

-

-	int letterdiff = abs( len1 - len2 );

-	// More than three letters different, don't bother

-	if ( letterdiff > 5 )

-		return false;

-

-	// Compute a "delta"

-	char *p1 = (char *)w1;

-	char *p2 = (char *)w2;

-

-	CUtlVector <char> word1;

-	CUtlVector <char> word2;

-

-	while ( *p1 )

-	{

-		if ( V_isalpha( *p1 ) )

-		{

-			word1.AddToTail( *p1 );

-		}

-		p1++;

-	}

-

-	while ( *p2 )

-	{

-		if ( V_isalpha( *p2 ) )

-		{

-			word2.AddToTail( *p2 );

-		}

-		p2++;

-	}

-

-	int i;

-	for ( i = 0; i < word1.Size(); i++ )

-	{

-		char c = word1[ i ];

-

-		// See if c is in word 2, if so subtract it out

-		int idx = word2.Find( c );

-

-		if ( idx != word2.InvalidIndex() )

-		{

-			word2.Remove( idx );

-		}

-	}

-

-	if ( word2.Size() <= letterdiff )

-		return true;

-

-	word2.RemoveAll();

-

-	while ( *p2 )

-	{

-		if ( V_isalpha( *p2 ) )

-		{

-			word2.AddToTail( *p2 );

-		}

-		p2++;

-	}

-

-	for ( i = 0; i < word2.Size(); i++ )

-	{

-		char c = word2[ i ];

-

-		// See if c is in word 2, if so subtract it out

-		int idx = word1.Find( c );

-

-		if ( idx != word1.InvalidIndex() )

-		{

-			word1.Remove( idx );

-		}

-	}

-

-	if ( word1.Size() <= letterdiff )

-		return true;

-

-	return false;

-}

-

-//-----------------------------------------------------------------------------

-// Purpose: For foreign language stuff, if inwords is empty, process anyway...

-// Input  : *analysis - 

-//			outwords - 

-//-----------------------------------------------------------------------------

-void CPhonemeExtractorLipSinc::ProcessWordsTextless( TALKBACK_ANALYSIS *analysis, CSentence& outwords )

-{

-	long count;

-

-	TALKBACK_ERR err = talkback->TalkBackGetNumPhonemes( analysis, &count );

-	if ( err != TALKBACK_NOERR )

-	{

-		DescribeError( err );

-		return;

-	}

-

-	CWordTag *newWord = new CWordTag;

-

-	newWord->SetWord( TEXTLESS_WORDNAME );

-

-	float starttime = 0.0f;

-	float endtime = 1.0f;

-

-

-	for ( int i = 0; i < count; ++i )

-	{

-		// Get phoneme and timing info

-		CAnalyzedPhoneme *ph = GetAnalyzedPhoneme( analysis, i );

-		if ( !ph )

-			continue;

-

-		CPhonemeTag *ptag = new CPhonemeTag;

-

-		if ( i == 0 || ( ph->starttime < starttime ) )

-		{

-			starttime = ph->starttime;

-		}

-

-		if ( i == 0 || ( ph->endtime > endtime ) )

-		{

-			endtime = ph->endtime;

-		}

-

-		ptag->SetStartTime( ph->starttime );

-		ptag->SetEndTime( ph->endtime );

-

-		ptag->m_uiStartByte = ComputeByteFromTime( ph->starttime );

-		ptag->m_uiEndByte = ComputeByteFromTime( ph->endtime );

-

-		ptag->SetTag( ph->phoneme );

-		ptag->SetPhonemeCode( TextToPhoneme( ptag->GetTag() ) );

-

-		newWord->m_Phonemes.AddToTail( ptag );

-	}

-

-	newWord->m_flStartTime = starttime;

-	newWord->m_flEndTime = endtime;

-

-	newWord->m_uiStartByte = ComputeByteFromTime( starttime );

-	newWord->m_uiEndByte = ComputeByteFromTime( endtime );

-

-	outwords.Reset();

-	outwords.AddWordTag( newWord );

-	outwords.SetTextFromWords();

-}

-

-//-----------------------------------------------------------------------------

-// Purpose: 

-// Input  : *analysis - 

-//			inwords - 

-//			outwords - 

-//-----------------------------------------------------------------------------

-void CPhonemeExtractorLipSinc::ProcessWords( TALKBACK_ANALYSIS *analysis, CSentence& inwords, CSentence& outwords )

-{

-	long count;

-

-	TALKBACK_ERR err = talkback->TalkBackGetNumWords( analysis, &count );

-	if ( err != TALKBACK_NOERR )

-	{

-		DescribeError( err );

-		return;

-	}

-

-	if ( count <= 0L )

-	{

-		if ( inwords.m_Words.Count() == 0 || 

-			!Q_strnicmp( inwords.GetText(), TEXTLESS_WORDNAME, Q_strlen( TEXTLESS_WORDNAME ) ) )

-		{

-			ProcessWordsTextless( analysis, outwords );

-		}

-		return;

-	}

-

-	// Bogus

-	if ( count >= 100000L )

-		return;

-

-	int inwordpos = 0;

-	int awordpos = 0;

-

-	outwords.Reset();

-

-	char previous[ 256 ];

-	previous[ 0 ] = 0;

-

-	while ( inwordpos < inwords.m_Words.Size() )

-	{

-		CWordTag *in = inwords.m_Words[ inwordpos ];

-

-		if ( awordpos >= count )

-		{

-			// Just copy the rest over without phonemes

-			CWordTag *copy = new CWordTag( *in );

-

-			outwords.AddWordTag( copy );

-

-			inwordpos++;

-			continue;

-		}

-

-		// Should never fail

-		CAnalyzedWord *w = GetAnalyzedWord( analysis, awordpos );

-		if ( !w )

-		{

-			return;

-		}

-

-		if ( !stricmp( w->buffer, "<SIL>" ) )

-		{

-			awordpos++;

-			continue;

-		}

-

-		char const *check = ApplyTBWordRules( in->GetWord() );

-		if ( !FuzzyWordMatch( check, w->buffer ) )

-		{

-			bool advance_input = true;

-			if ( previous[ 0 ] )

-			{

-				if ( FuzzyWordMatch( previous, w->buffer ) )

-				{

-					advance_input = false;

-				}

-			}

-

-			if ( advance_input )

-			{

-				inwordpos++;

-			}

-			awordpos++;

-			continue;

-		}

-		strcpy( previous, check );

-

-		CWordTag *newWord = new CWordTag;

-

-		newWord->SetWord( in->GetWord() );

-

-		newWord->m_flStartTime = w->starttime;

-		newWord->m_flEndTime = w->endtime;

-

-		newWord->m_uiStartByte = ComputeByteFromTime( w->starttime );

-		newWord->m_uiEndByte = ComputeByteFromTime( w->endtime );

-

-		int phonemestart, phonemeend;

-

-		phonemestart = GetPhonemeIndexAtWordStart( analysis, w->starttime );

-		phonemeend = GetPhonemeIndexAtWordEnd( analysis, w->endtime );

-

-		if ( phonemestart >= 0 && phonemeend >= 0 )

-		{

-			for ( ; phonemestart <= phonemeend; phonemestart++ )

-			{

-				// Get phoneme and timing info

-				CAnalyzedPhoneme *ph = GetAnalyzedPhoneme( analysis, phonemestart );

-				if ( !ph )

-					continue;

-

-				CPhonemeTag *ptag = new CPhonemeTag;

-				ptag->SetStartTime( ph->starttime );

-				ptag->SetEndTime( ph->endtime );

-

-				ptag->m_uiStartByte = ComputeByteFromTime( ph->starttime );

-				ptag->m_uiEndByte = ComputeByteFromTime( ph->endtime );

-

-				ptag->SetTag( ph->phoneme );

-				ptag->SetPhonemeCode( TextToPhoneme( ptag->GetTag() ) );

-

-				newWord->m_Phonemes.AddToTail( ptag );

-			}

-		}

-

-		outwords.AddWordTag( newWord );

-		inwordpos++;

-		awordpos++;

-	}

-}

-

-char const *CPhonemeExtractorLipSinc::ApplyTBWordRules( char const *word )

-{

-	static char outword[ 256 ];

-

-	char const *in = word;

-	char *out = outword;

-

-	while ( *in && ( ( out - outword ) <= 255 ) )

-	{

-		if ( *in == '\t' ||

-			 *in == ' ' ||

-			 *in == '\n' ||

-			 *in == '-' ||

-			 *in == '.' ||

-			 *in == ',' ||

-			 *in == ';' ||

-			 *in == '?' ||

-			 *in == '"' ||

-			 *in == ':' ||

-			 *in == '(' ||

-			 *in == ')' )

-		{

-			in++;

-			*out++ = ' ';

-			continue;

-		}

-

-		if ( !V_isprint( *in ) )

-		{

-			in++;

-			continue;

-		}

-

-		if ( *in >= 128 )

-		{

-			in++;

-			continue;

-		}

-

-		// Skip numbers

-		if ( *in >= '0' && *in <= '9' )

-		{

-			in++;

-			continue;

-		}

-

-		// Convert all letters to upper case

-		if ( *in >= 'a' && *in <= 'z' )

-		{

-			*out++ = ( *in++ ) - 'a' + 'A';

-			continue;

-		}

-

-		if ( *in >= 'A' && *in <= 'Z' )

-		{

-			*out++ = *in++;

-			continue;

-		}

-

-		if ( *in == '\'' )

-		{

-			*out++ = *in++;

-			continue;

-		}

-

-		in++;

-	}

-

-	*out = 0;

-

-	return outword;

-}

-

-//-----------------------------------------------------------------------------

-// Purpose: Given a wavfile and a list of inwords, determines the word/phonene 

-//  sample counts for the sentce

-// Output : SR_RESULT

-//-----------------------------------------------------------------------------

-SR_RESULT CPhonemeExtractorLipSinc::Extract( 

-	const char *wavfile,

-	int numsamples,

-	void (*pfnPrint)( const char *fmt, ... ),

-	CSentence& inwords,

-	CSentence& outwords )

-{

-	// g_enableTalkBackDebuggingOutput = 1;

-

-	m_pfnPrint = pfnPrint;

-

-	if ( !InitLipSinc() )

-	{

-		return SR_RESULT_ERROR;

-	}

-	

-	m_flSampleCount = numsamples;

-

-	if ( !CheckSoundFile( wavfile ) )

-	{

-		FreeLibrary( m_hHelper );

-		return SR_RESULT_ERROR;

-	}

-

-	TALKBACK_ANALYSIS *analysis = NULL;

-

-	if ( !AttemptAnalysis( &analysis, wavfile, inwords ) )

-	{

-		FreeLibrary( m_hHelper );

-		return SR_RESULT_FAILED;

-	}

-	

-	if ( strlen( inwords.GetText() ) <= 0 )

-	{

-		inwords.SetTextFromWords();

-	}

-

-	outwords = inwords;

-

-	// Examine data

-	ProcessWords( analysis, inwords, outwords );

-

-	if ( analysis )

-	{

-		talkback->TalkBackFreeAnalysis( &analysis );

-	}

-

-	return SR_RESULT_SUCCESS;

-}

-

+//========= Copyright Valve Corporation, All rights reserved. ============//
+//
+// Purpose: 
+//
+// $NoKeywords: $
+//=============================================================================//
+#include <stdio.h>
+#include <stdarg.h>
+#include <memory.h>
+#include <windows.h>
+#include <mmsystem.h>
+#include <mmreg.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+
+#include "phonemeextractor/PhonemeExtractor.h"
+#include "ims_helper/ims_helper.h"
+
+#include "tier0/dbg.h"
+#include "sentence.h"
+#include "PhonemeConverter.h"
+#include "tier1/strtools.h"
+
+#define TEXTLESS_WORDNAME	"[Textless]"
+
+static IImsHelper *talkback = NULL;
+
+//-----------------------------------------------------------------------------
+// Purpose: Expose the interface
+//-----------------------------------------------------------------------------
+class CPhonemeExtractorLipSinc : public IPhonemeExtractor
+{
+public:
+	virtual PE_APITYPE	GetAPIType() const
+	{
+		return SPEECH_API_LIPSINC;
+	}
+
+	// Used for menus, etc
+	virtual char const *GetName() const
+	{
+		return "IMS (LipSinc)";
+	}
+
+	SR_RESULT Extract( 
+		const char *wavfile,
+		int numsamples,
+		void (*pfnPrint)( const char *fmt, ... ),
+		CSentence& inwords,
+		CSentence& outwords );
+
+
+	CPhonemeExtractorLipSinc( void );
+	~CPhonemeExtractorLipSinc( void );
+
+	enum
+	{
+		MAX_WORD_LENGTH = 128,
+	};
+private:
+
+
+	class CAnalyzedWord
+	{
+	public:
+		char		buffer[ MAX_WORD_LENGTH ];
+		double		starttime;
+		double		endtime;
+	};
+
+	class CAnalyzedPhoneme
+	{
+	public:
+		char		phoneme[ 32 ];
+		double		starttime;
+		double		endtime;
+	};
+
+	bool InitLipSinc( void );
+	void ShutdownLipSinc( void );
+
+	void DescribeError( TALKBACK_ERR err );
+	void Printf( char const *fmt, ... );
+
+	bool CheckSoundFile( char const *filename );
+	bool GetInitialized( void );
+	void SetInitialized( bool init );
+
+	void (*m_pfnPrint)( const char *fmt, ... );
+
+	char const *ConstructInputSentence( CSentence& inwords );
+	bool AttemptAnalysis( TALKBACK_ANALYSIS **ppAnalysis, char const *wavfile, CSentence& inwords );
+
+	char const *ApplyTBWordRules( char const *word );
+
+	void ProcessWords( TALKBACK_ANALYSIS *analysis, CSentence& inwords, CSentence& outwords );
+	void ProcessWordsTextless( TALKBACK_ANALYSIS *analysis, CSentence& outwords );
+
+	int GetPhonemeIndexAtWord( TALKBACK_ANALYSIS *analysis, double time, bool checkstart );
+
+	int GetPhonemeIndexAtWordStart( TALKBACK_ANALYSIS *analysis, double starttime );
+	int GetPhonemeIndexAtWordEnd( TALKBACK_ANALYSIS *analysis, double endtime );
+
+	CAnalyzedWord *GetAnalyzedWord( TALKBACK_ANALYSIS *analysis, int index );
+	CAnalyzedPhoneme *GetAnalyzedPhoneme( TALKBACK_ANALYSIS *analysis, int index );
+
+	int ComputeByteFromTime( float time );
+
+	bool m_bInitialized;
+
+	float	m_flSampleCount;
+	float	m_flDuration;
+
+	float	m_flSamplesPerSecond;
+
+	int		m_nBytesPerSample;
+
+	HMODULE m_hHelper;
+};
+
+CPhonemeExtractorLipSinc::CPhonemeExtractorLipSinc( void )
+{
+	m_hHelper = (HMODULE)0;
+	m_pfnPrint = NULL;
+
+	m_bInitialized = false;
+	
+	m_flSampleCount = 0.0f;
+	m_flDuration = 0.0f;
+
+	m_flSamplesPerSecond = 0.0f;
+
+	m_nBytesPerSample = 0;
+}
+
+CPhonemeExtractorLipSinc::~CPhonemeExtractorLipSinc( void )
+{
+	if ( GetInitialized() )
+	{
+		ShutdownLipSinc();
+	}
+}
+
+bool CPhonemeExtractorLipSinc::GetInitialized( void )
+{
+	return m_bInitialized;
+}
+
+void CPhonemeExtractorLipSinc::SetInitialized( bool init )
+{
+	m_bInitialized = init;
+}
+
+int CPhonemeExtractorLipSinc::ComputeByteFromTime( float time )
+{
+	if ( !m_flDuration )
+		return 0;
+
+	float frac = time / m_flDuration;
+
+	float sampleNumber = frac * m_flSampleCount;
+
+	int bytenumber = sampleNumber * m_nBytesPerSample;
+
+	return bytenumber;
+}
+
+void CPhonemeExtractorLipSinc::DescribeError( TALKBACK_ERR err )
+{
+	Assert( m_pfnPrint );
+
+	// Get the error description.
+	char errorDesc[256] = "";
+	if ( err != TALKBACK_NOERR )
+	{
+		talkback->TalkBackGetErrorString( err, sizeof(errorDesc), errorDesc );
+	}
+	
+	// Report or log the error...
+	(*m_pfnPrint)( "LIPSINC ERROR:  %s\n", errorDesc );
+}
+
+//-----------------------------------------------------------------------------
+// Purpose: 
+// Input  : *fmt - 
+//			.. - 
+//-----------------------------------------------------------------------------
+void CPhonemeExtractorLipSinc::Printf( char const *fmt, ... )
+{
+	Assert( m_pfnPrint );
+
+	char string[ 4096 ];
+
+	va_list argptr;
+	va_start( argptr, fmt );
+	vsprintf( string, fmt, argptr );
+	va_end( argptr );
+
+	(*m_pfnPrint)( "%s", string );
+}
+
+bool CPhonemeExtractorLipSinc::CheckSoundFile( char const *filename )
+{
+	TALKBACK_SOUND_FILE_METRICS fm;
+	memset( &fm, 0, sizeof( fm ) );
+	fm.m_size = sizeof( fm );
+
+	TALKBACK_ERR err = talkback->TalkBackGetSoundFileMetrics( filename, &fm );
+	if ( err != TALKBACK_NOERR )
+	{
+		DescribeError( err );
+		return false;
+	}
+
+	if ( fm.m_canBeAnalyzed )
+	{
+		Printf( "%s:  %.2f s, rate %i, bits %i, channels %i\n",
+			filename,
+			fm.m_duration,
+			fm.m_sampleRate,
+			fm.m_bitsPerSample,
+			fm.m_channelCount );
+	}
+
+	m_flDuration = fm.m_duration;
+	if ( m_flDuration > 0 )
+	{
+		m_flSamplesPerSecond = m_flSampleCount / m_flDuration;
+	}
+	else
+	{
+		m_flSamplesPerSecond = 0.0f;
+	}
+
+	m_nBytesPerSample = ( fm.m_bitsPerSample >> 3 );
+
+	m_flSampleCount /= m_nBytesPerSample;
+
+	m_nBytesPerSample /= fm.m_channelCount;
+
+	return fm.m_canBeAnalyzed ? true : false;
+}
+
+typedef IImsHelper *(*pfnImsHelper)(void);
+
+//-----------------------------------------------------------------------------
+// Purpose: 
+// Output : Returns true on success, false on failure.
+//-----------------------------------------------------------------------------
+bool CPhonemeExtractorLipSinc::InitLipSinc( void )
+{
+	if ( GetInitialized() )
+	{
+		return true;
+	}
+
+	m_hHelper = LoadLibrary( "ims_helper.dll" );
+	if ( !m_hHelper )
+	{
+		return false;
+	}
+	
+	pfnImsHelper factory = (pfnImsHelper)::GetProcAddress( m_hHelper, "GetImsHelper" );
+	if ( !factory )
+	{
+		FreeLibrary( m_hHelper );
+		return false;
+	}
+
+	talkback = reinterpret_cast< IImsHelper * >( (*factory)() );
+	if ( !talkback )
+	{
+		FreeLibrary( m_hHelper );
+		return false;
+	}
+
+	char szExeName[ MAX_PATH ];
+	szExeName[0] = 0;
+	GetModuleFileName( (HMODULE)0, szExeName, sizeof( szExeName ) );
+
+	char szBaseDir[ MAX_PATH ];
+	Q_strncpy( szBaseDir, szExeName, sizeof( szBaseDir ) );
+
+	Q_StripLastDir( szBaseDir, sizeof( szBaseDir ) );
+	Q_StripTrailingSlash( szBaseDir );
+	Q_strlower( szBaseDir );
+
+	char coreDataDir[ 512 ];
+	Q_snprintf( coreDataDir, sizeof( coreDataDir ), "%s\\lipsinc_data\\",
+		szBaseDir );
+	Q_FixSlashes( coreDataDir );
+
+	char szCheck[ 512 ];
+	Q_snprintf( szCheck, sizeof( szCheck ), "%sDtC6dal.dat", coreDataDir );
+	struct __stat64 buf;
+
+	if ( _stat64( szCheck, &buf ) != 0 )
+	{
+		Q_snprintf( coreDataDir, sizeof( coreDataDir ), "%s\\bin\\lipsinc_data\\",
+			szBaseDir );
+		Q_FixSlashes( coreDataDir );
+		Q_snprintf( szCheck, sizeof( szCheck ), "%sDtC6dal.dat", coreDataDir );
+
+		if ( _stat64( szCheck, &buf ) != 0 )
+		{
+			Error( "Unable to find talkback data files in %s.", coreDataDir );
+		}
+	}
+
+	TALKBACK_ERR err;
+	
+	err = talkback->TalkBackStartupLibrary( coreDataDir );
+	if ( err != TALKBACK_NOERR )
+	{
+		DescribeError( err );
+		FreeLibrary( m_hHelper );
+		return false;
+	}
+
+	long verMajor = 0;
+	long verMinor = 0;
+	long verRevision = 0;
+	
+	err = talkback->TalkBackGetVersion(
+		&verMajor, 
+		&verMinor, 
+		&verRevision);
+	if ( err != TALKBACK_NOERR )
+	{
+		DescribeError( err );
+		FreeLibrary( m_hHelper );
+		return false;
+	}
+
+	Printf( "Lipsinc TalkBack Version %i.%i.%i\n", verMajor, verMinor, verRevision );
+
+	m_bInitialized = true;
+
+	return true;
+}
+
+//-----------------------------------------------------------------------------
+// Purpose: 
+//-----------------------------------------------------------------------------
+void CPhonemeExtractorLipSinc::ShutdownLipSinc( void )
+{
+	// HACK HACK:  This seems to crash on exit sometimes
+	__try
+	{
+		talkback->TalkBackShutdownLibrary();
+
+		FreeLibrary( m_hHelper );
+	}
+	__except(EXCEPTION_EXECUTE_HANDLER )
+	{
+		OutputDebugString( "----> Crash shutting down TALKBACK sdk, exception caught and ignored\n" );
+	}
+}
+
+//-----------------------------------------------------------------------------
+// Purpose: 
+// Input  : inwords - 
+// Output : char const
+//-----------------------------------------------------------------------------
+char const *CPhonemeExtractorLipSinc::ConstructInputSentence( CSentence& inwords )
+{
+	static char sentence[ 16384 ];
+
+	sentence[ 0 ] = 0;
+
+	int last = inwords.m_Words.Size() - 1;
+
+	for ( int i = 0 ; i <= last; i++ )
+	{
+		CWordTag *w = inwords.m_Words[ i ];
+
+		strcat( sentence, w->GetWord() );
+		if ( i != last )
+		{
+			strcat( sentence, " " );
+		}
+	}
+
+	if ( inwords.m_Words.Count() == 1 && 
+		!Q_strnicmp( inwords.GetText(), TEXTLESS_WORDNAME, Q_strlen( TEXTLESS_WORDNAME ) ) )
+	{
+		sentence[ 0 ] = 0;
+	}
+
+	return sentence;
+}
+
+bool CPhonemeExtractorLipSinc::AttemptAnalysis( TALKBACK_ANALYSIS **ppAnalysis, char const *wavfile, CSentence& inwords )
+{
+	*ppAnalysis = NULL;
+
+	TALKBACK_ANALYSIS_SETTINGS settings;
+	memset( &settings, 0, sizeof( settings ) );
+
+   // Set this field to sizeof(TALKBACK_ANALYSIS_SETTINGS) before using the
+    // structure.
+	settings.fSize = sizeof( TALKBACK_ANALYSIS_SETTINGS );
+	
+
+    // Default value: 30 (frames per second).
+    settings.fFrameRate = 100;
+    // Set this to 1 to optimize for flipbook output, 0 to do analysis normally.
+    //
+    // Default value: 0 (normal analysis).
+    settings.fOptimizeForFlipbook = 0;
+    // Set this to -1 to seed the random number generator with the current time.
+    // Any other number will be used directly for the random number seed, which
+    // is useful if you want repeatable speech gestures. This value does not
+    // influence lip-synching at all.
+    //
+    // Default value: -1 (use current time).
+    settings.fRandomSeed = -1;
+    // Path to the configuration (.INI) file with phoneme-to-speech-target
+    // mapping. Set this to NULL to use the default mapping.
+    //
+    // Default value: NULL (use default mapping).
+    settings.fConfigFile = NULL;
+
+	char const *text = ConstructInputSentence( inwords );
+
+	Printf( "Analyzing: \"%s\"\n", text[ 0 ] ? text : TEXTLESS_WORDNAME );
+
+	TALKBACK_ERR err = talkback->TalkBackGetAnalysis( 
+		ppAnalysis,
+		wavfile,
+		text,
+		&settings );
+
+	if ( err != TALKBACK_NOERR )
+	{
+		DescribeError( err );
+		return false;
+	}
+
+	Printf( "Analysis successful...\n" );
+
+	return true;
+}
+
+typedef struct
+{
+	TALKBACK_PHONEME phoneme;
+	char const		*string;
+} TBPHONEMES_t;
+
+static TBPHONEMES_t g_TBPhonemeList[]=
+{
+	{ TALKBACK_PHONEME_IY, "iy" },
+	{ TALKBACK_PHONEME_IH, "ih" },
+	{ TALKBACK_PHONEME_EH, "eh" },
+	{ TALKBACK_PHONEME_EY, "ey" },
+	{ TALKBACK_PHONEME_AE, "ae" },
+	{ TALKBACK_PHONEME_AA, "aa" },
+	{ TALKBACK_PHONEME_AW, "aw" },
+	{ TALKBACK_PHONEME_AY, "ay" },
+	{ TALKBACK_PHONEME_AH, "ah" },
+	{ TALKBACK_PHONEME_AO, "ao" },
+	{ TALKBACK_PHONEME_OY, "oy" },
+	{ TALKBACK_PHONEME_OW, "ow" },
+	{ TALKBACK_PHONEME_UH, "uh" },
+	{ TALKBACK_PHONEME_UW, "uw" },
+	{ TALKBACK_PHONEME_ER, "er" },
+	{ TALKBACK_PHONEME_AX, "ax" },
+	{ TALKBACK_PHONEME_S, "s" },
+	{ TALKBACK_PHONEME_SH, "sh" },
+	{ TALKBACK_PHONEME_Z, "z" },
+	{ TALKBACK_PHONEME_ZH, "zh" },
+	{ TALKBACK_PHONEME_F, "f" },
+	{ TALKBACK_PHONEME_TH, "th" },
+	{ TALKBACK_PHONEME_V, "v" },
+	{ TALKBACK_PHONEME_DH, "dh" },
+	{ TALKBACK_PHONEME_M, "m" },
+	{ TALKBACK_PHONEME_N, "n" },
+	{ TALKBACK_PHONEME_NG, "ng" },
+	{ TALKBACK_PHONEME_L, "l" },
+	{ TALKBACK_PHONEME_R, "r" },
+	{ TALKBACK_PHONEME_W, "w" },
+	{ TALKBACK_PHONEME_Y, "y" },
+	{ TALKBACK_PHONEME_HH, "hh" },
+	{ TALKBACK_PHONEME_B, "b" },
+	{ TALKBACK_PHONEME_D, "d" },
+	{ TALKBACK_PHONEME_JH, "jh" },
+	{ TALKBACK_PHONEME_G, "g" },
+	{ TALKBACK_PHONEME_P, "p" },
+	{ TALKBACK_PHONEME_T, "t" },
+	{ TALKBACK_PHONEME_K, "k" },
+	{ TALKBACK_PHONEME_CH, "ch" },
+	{ TALKBACK_PHONEME_SIL, "<sil>" },
+	{ -1, NULL }
+};
+
+char const *TBPhonemeToString( TALKBACK_PHONEME phoneme )
+{
+	if ( phoneme < TALKBACK_PHONEME_FIRST || phoneme > TALKBACK_PHONEME_LAST )
+	{
+		return "Bogus";
+	}
+
+	TBPHONEMES_t *item = &g_TBPhonemeList[ phoneme ];
+	return item->string;
+}
+
+//-----------------------------------------------------------------------------
+// Purpose: 
+// Input  : *analysis - 
+//			time - 
+//			start - 
+// Output : int
+//-----------------------------------------------------------------------------
+int CPhonemeExtractorLipSinc::GetPhonemeIndexAtWord( TALKBACK_ANALYSIS *analysis, double time, bool start )
+{
+	long count;
+
+	TALKBACK_ERR err = talkback->TalkBackGetNumPhonemes( analysis, &count );
+	if ( err != TALKBACK_NOERR )
+	{
+		DescribeError( err );
+		return -1;
+	}
+
+	if ( count <= 0L )
+		return -1;
+
+	// Bogus
+	if ( count >= 100000L )
+		return -1;
+
+	for ( int i = 0; i < (int)count; i++ )
+	{
+		TALKBACK_PHONEME tbPhoneme = TALKBACK_PHONEME_INVALID;
+		err = talkback->TalkBackGetPhonemeEnum( analysis, i, &tbPhoneme );
+		if ( err != TALKBACK_NOERR )
+		{
+			DescribeError( err );
+			continue;
+		}
+
+		double t;
+
+		if ( start )
+		{
+			err = talkback->TalkBackGetPhonemeStartTime( analysis, i, &t );
+		}
+		else
+		{
+			err = talkback->TalkBackGetPhonemeEndTime( analysis, i, &t );
+		}
+
+		if ( err != TALKBACK_NOERR )
+		{
+			DescribeError( err );
+			continue;
+		}
+
+		if ( t == time )
+		{
+			return i;
+		}
+	}
+
+	return -1;
+}
+
+//-----------------------------------------------------------------------------
+// Purpose: 
+// Input  : *analysis - 
+//			starttime - 
+// Output : int
+//-----------------------------------------------------------------------------
+int CPhonemeExtractorLipSinc::GetPhonemeIndexAtWordStart( TALKBACK_ANALYSIS *analysis, double starttime )
+{
+	return GetPhonemeIndexAtWord( analysis, starttime, true );
+}
+
+//-----------------------------------------------------------------------------
+// Purpose: 
+// Input  : *analysis - 
+//			endtime - 
+// Output : int
+//-----------------------------------------------------------------------------
+int CPhonemeExtractorLipSinc::GetPhonemeIndexAtWordEnd( TALKBACK_ANALYSIS *analysis, double endtime )
+{
+	return GetPhonemeIndexAtWord( analysis, endtime, false );
+}
+
+CPhonemeExtractorLipSinc::CAnalyzedPhoneme *CPhonemeExtractorLipSinc::GetAnalyzedPhoneme( TALKBACK_ANALYSIS *analysis, int index )
+{
+	static CAnalyzedPhoneme p;
+
+	memset( &p, 0, sizeof( p ) );
+
+	TALKBACK_PHONEME tb;
+
+	TALKBACK_ERR err = talkback->TalkBackGetPhonemeEnum( analysis, index, &tb );
+	if ( err != TALKBACK_NOERR )
+	{
+		DescribeError( err );
+		return NULL;
+	}
+
+	strcpy( p.phoneme, TBPhonemeToString( tb ) );
+
+	err = talkback->TalkBackGetPhonemeStartTime( analysis, index, &p.starttime );
+	if ( err != TALKBACK_NOERR )
+	{
+		DescribeError( err );
+		return NULL;
+	}
+	err = talkback->TalkBackGetPhonemeEndTime( analysis, index, &p.endtime );
+	if ( err != TALKBACK_NOERR )
+	{
+		DescribeError( err );
+		return NULL;
+	}
+
+	return &p;
+}
+
+CPhonemeExtractorLipSinc::CAnalyzedWord *CPhonemeExtractorLipSinc::GetAnalyzedWord( TALKBACK_ANALYSIS *analysis, int index )
+{
+	static CAnalyzedWord w;
+
+	memset( &w, 0, sizeof( w ) );
+
+	long chars = sizeof( w.buffer );
+
+	TALKBACK_ERR err = talkback->TalkBackGetWord( analysis, index, chars, w.buffer );
+	if ( err != TALKBACK_NOERR )
+	{
+		DescribeError( err );
+		return NULL;
+	}
+
+	err = talkback->TalkBackGetWordStartTime( analysis, index, &w.starttime );
+	if ( err != TALKBACK_NOERR )
+	{
+		DescribeError( err );
+		return NULL;
+	}
+	err = talkback->TalkBackGetWordEndTime( analysis, index, &w.endtime );
+	if ( err != TALKBACK_NOERR )
+	{
+		DescribeError( err );
+		return NULL;
+	}
+
+	return &w;
+}
+
+//-----------------------------------------------------------------------------
+// Purpose: 
+// Input  : *w1 - 
+//			*w2 - 
+// Output : Returns true on success, false on failure.
+//-----------------------------------------------------------------------------
+bool FuzzyWordMatch( char const *w1, char const *w2 )
+{
+	int len1 = strlen( w1 );
+	int len2 = strlen( w2 );
+
+	int minlen = min( len1, len2 );
+
+	// Found a match
+	if ( !strnicmp( w1, w2, minlen ) )
+		return true;
+
+	int letterdiff = abs( len1 - len2 );
+	// More than three letters different, don't bother
+	if ( letterdiff > 5 )
+		return false;
+
+	// Compute a "delta"
+	char *p1 = (char *)w1;
+	char *p2 = (char *)w2;
+
+	CUtlVector <char> word1;
+	CUtlVector <char> word2;
+
+	while ( *p1 )
+	{
+		if ( V_isalpha( *p1 ) )
+		{
+			word1.AddToTail( *p1 );
+		}
+		p1++;
+	}
+
+	while ( *p2 )
+	{
+		if ( V_isalpha( *p2 ) )
+		{
+			word2.AddToTail( *p2 );
+		}
+		p2++;
+	}
+
+	int i;
+	for ( i = 0; i < word1.Size(); i++ )
+	{
+		char c = word1[ i ];
+
+		// See if c is in word 2, if so subtract it out
+		int idx = word2.Find( c );
+
+		if ( idx != word2.InvalidIndex() )
+		{
+			word2.Remove( idx );
+		}
+	}
+
+	if ( word2.Size() <= letterdiff )
+		return true;
+
+	word2.RemoveAll();
+
+	while ( *p2 )
+	{
+		if ( V_isalpha( *p2 ) )
+		{
+			word2.AddToTail( *p2 );
+		}
+		p2++;
+	}
+
+	for ( i = 0; i < word2.Size(); i++ )
+	{
+		char c = word2[ i ];
+
+		// See if c is in word 2, if so subtract it out
+		int idx = word1.Find( c );
+
+		if ( idx != word1.InvalidIndex() )
+		{
+			word1.Remove( idx );
+		}
+	}
+
+	if ( word1.Size() <= letterdiff )
+		return true;
+
+	return false;
+}
+
+//-----------------------------------------------------------------------------
+// Purpose: For foreign language stuff, if inwords is empty, process anyway...
+// Input  : *analysis - 
+//			outwords - 
+//-----------------------------------------------------------------------------
+void CPhonemeExtractorLipSinc::ProcessWordsTextless( TALKBACK_ANALYSIS *analysis, CSentence& outwords )
+{
+	long count;
+
+	TALKBACK_ERR err = talkback->TalkBackGetNumPhonemes( analysis, &count );
+	if ( err != TALKBACK_NOERR )
+	{
+		DescribeError( err );
+		return;
+	}
+
+	CWordTag *newWord = new CWordTag;
+
+	newWord->SetWord( TEXTLESS_WORDNAME );
+
+	float starttime = 0.0f;
+	float endtime = 1.0f;
+
+
+	for ( int i = 0; i < count; ++i )
+	{
+		// Get phoneme and timing info
+		CAnalyzedPhoneme *ph = GetAnalyzedPhoneme( analysis, i );
+		if ( !ph )
+			continue;
+
+		CPhonemeTag *ptag = new CPhonemeTag;
+
+		if ( i == 0 || ( ph->starttime < starttime ) )
+		{
+			starttime = ph->starttime;
+		}
+
+		if ( i == 0 || ( ph->endtime > endtime ) )
+		{
+			endtime = ph->endtime;
+		}
+
+		ptag->SetStartTime( ph->starttime );
+		ptag->SetEndTime( ph->endtime );
+
+		ptag->m_uiStartByte = ComputeByteFromTime( ph->starttime );
+		ptag->m_uiEndByte = ComputeByteFromTime( ph->endtime );
+
+		ptag->SetTag( ph->phoneme );
+		ptag->SetPhonemeCode( TextToPhoneme( ptag->GetTag() ) );
+
+		newWord->m_Phonemes.AddToTail( ptag );
+	}
+
+	newWord->m_flStartTime = starttime;
+	newWord->m_flEndTime = endtime;
+
+	newWord->m_uiStartByte = ComputeByteFromTime( starttime );
+	newWord->m_uiEndByte = ComputeByteFromTime( endtime );
+
+	outwords.Reset();
+	outwords.AddWordTag( newWord );
+	outwords.SetTextFromWords();
+}
+
+//-----------------------------------------------------------------------------
+// Purpose: 
+// Input  : *analysis - 
+//			inwords - 
+//			outwords - 
+//-----------------------------------------------------------------------------
+void CPhonemeExtractorLipSinc::ProcessWords( TALKBACK_ANALYSIS *analysis, CSentence& inwords, CSentence& outwords )
+{
+	long count;
+
+	TALKBACK_ERR err = talkback->TalkBackGetNumWords( analysis, &count );
+	if ( err != TALKBACK_NOERR )
+	{
+		DescribeError( err );
+		return;
+	}
+
+	if ( count <= 0L )
+	{
+		if ( inwords.m_Words.Count() == 0 || 
+			!Q_strnicmp( inwords.GetText(), TEXTLESS_WORDNAME, Q_strlen( TEXTLESS_WORDNAME ) ) )
+		{
+			ProcessWordsTextless( analysis, outwords );
+		}
+		return;
+	}
+
+	// Bogus
+	if ( count >= 100000L )
+		return;
+
+	int inwordpos = 0;
+	int awordpos = 0;
+
+	outwords.Reset();
+
+	char previous[ 256 ];
+	previous[ 0 ] = 0;
+
+	while ( inwordpos < inwords.m_Words.Size() )
+	{
+		CWordTag *in = inwords.m_Words[ inwordpos ];
+
+		if ( awordpos >= count )
+		{
+			// Just copy the rest over without phonemes
+			CWordTag *copy = new CWordTag( *in );
+
+			outwords.AddWordTag( copy );
+
+			inwordpos++;
+			continue;
+		}
+
+		// Should never fail
+		CAnalyzedWord *w = GetAnalyzedWord( analysis, awordpos );
+		if ( !w )
+		{
+			return;
+		}
+
+		if ( !stricmp( w->buffer, "<SIL>" ) )
+		{
+			awordpos++;
+			continue;
+		}
+
+		char const *check = ApplyTBWordRules( in->GetWord() );
+		if ( !FuzzyWordMatch( check, w->buffer ) )
+		{
+			bool advance_input = true;
+			if ( previous[ 0 ] )
+			{
+				if ( FuzzyWordMatch( previous, w->buffer ) )
+				{
+					advance_input = false;
+				}
+			}
+
+			if ( advance_input )
+			{
+				inwordpos++;
+			}
+			awordpos++;
+			continue;
+		}
+		strcpy( previous, check );
+
+		CWordTag *newWord = new CWordTag;
+
+		newWord->SetWord( in->GetWord() );
+
+		newWord->m_flStartTime = w->starttime;
+		newWord->m_flEndTime = w->endtime;
+
+		newWord->m_uiStartByte = ComputeByteFromTime( w->starttime );
+		newWord->m_uiEndByte = ComputeByteFromTime( w->endtime );
+
+		int phonemestart, phonemeend;
+
+		phonemestart = GetPhonemeIndexAtWordStart( analysis, w->starttime );
+		phonemeend = GetPhonemeIndexAtWordEnd( analysis, w->endtime );
+
+		if ( phonemestart >= 0 && phonemeend >= 0 )
+		{
+			for ( ; phonemestart <= phonemeend; phonemestart++ )
+			{
+				// Get phoneme and timing info
+				CAnalyzedPhoneme *ph = GetAnalyzedPhoneme( analysis, phonemestart );
+				if ( !ph )
+					continue;
+
+				CPhonemeTag *ptag = new CPhonemeTag;
+				ptag->SetStartTime( ph->starttime );
+				ptag->SetEndTime( ph->endtime );
+
+				ptag->m_uiStartByte = ComputeByteFromTime( ph->starttime );
+				ptag->m_uiEndByte = ComputeByteFromTime( ph->endtime );
+
+				ptag->SetTag( ph->phoneme );
+				ptag->SetPhonemeCode( TextToPhoneme( ptag->GetTag() ) );
+
+				newWord->m_Phonemes.AddToTail( ptag );
+			}
+		}
+
+		outwords.AddWordTag( newWord );
+		inwordpos++;
+		awordpos++;
+	}
+}
+
+char const *CPhonemeExtractorLipSinc::ApplyTBWordRules( char const *word )
+{
+	static char outword[ 256 ];
+
+	char const *in = word;
+	char *out = outword;
+
+	while ( *in && ( ( out - outword ) <= 255 ) )
+	{
+		if ( *in == '\t' ||
+			 *in == ' ' ||
+			 *in == '\n' ||
+			 *in == '-' ||
+			 *in == '.' ||
+			 *in == ',' ||
+			 *in == ';' ||
+			 *in == '?' ||
+			 *in == '"' ||
+			 *in == ':' ||
+			 *in == '(' ||
+			 *in == ')' )
+		{
+			in++;
+			*out++ = ' ';
+			continue;
+		}
+
+		if ( !V_isprint( *in ) )
+		{
+			in++;
+			continue;
+		}
+
+		if ( *in >= 128 )
+		{
+			in++;
+			continue;
+		}
+
+		// Skip numbers
+		if ( *in >= '0' && *in <= '9' )
+		{
+			in++;
+			continue;
+		}
+
+		// Convert all letters to upper case
+		if ( *in >= 'a' && *in <= 'z' )
+		{
+			*out++ = ( *in++ ) - 'a' + 'A';
+			continue;
+		}
+
+		if ( *in >= 'A' && *in <= 'Z' )
+		{
+			*out++ = *in++;
+			continue;
+		}
+
+		if ( *in == '\'' )
+		{
+			*out++ = *in++;
+			continue;
+		}
+
+		in++;
+	}
+
+	*out = 0;
+
+	return outword;
+}
+
+//-----------------------------------------------------------------------------
+// Purpose: Given a wavfile and a list of inwords, determines the word/phonene 
+//  sample counts for the sentce
+// Output : SR_RESULT
+//-----------------------------------------------------------------------------
+SR_RESULT CPhonemeExtractorLipSinc::Extract( 
+	const char *wavfile,
+	int numsamples,
+	void (*pfnPrint)( const char *fmt, ... ),
+	CSentence& inwords,
+	CSentence& outwords )
+{
+	// g_enableTalkBackDebuggingOutput = 1;
+
+	m_pfnPrint = pfnPrint;
+
+	if ( !InitLipSinc() )
+	{
+		return SR_RESULT_ERROR;
+	}
+	
+	m_flSampleCount = numsamples;
+
+	if ( !CheckSoundFile( wavfile ) )
+	{
+		FreeLibrary( m_hHelper );
+		return SR_RESULT_ERROR;
+	}
+
+	TALKBACK_ANALYSIS *analysis = NULL;
+
+	if ( !AttemptAnalysis( &analysis, wavfile, inwords ) )
+	{
+		FreeLibrary( m_hHelper );
+		return SR_RESULT_FAILED;
+	}
+	
+	if ( strlen( inwords.GetText() ) <= 0 )
+	{
+		inwords.SetTextFromWords();
+	}
+
+	outwords = inwords;
+
+	// Examine data
+	ProcessWords( analysis, inwords, outwords );
+
+	if ( analysis )
+	{
+		talkback->TalkBackFreeAnalysis( &analysis );
+	}
+
+	return SR_RESULT_SUCCESS;
+}
+
 EXPOSE_SINGLE_INTERFACE( CPhonemeExtractorLipSinc, IPhonemeExtractor, VPHONEME_EXTRACTOR_INTERFACE );
\ No newline at end of file
diff --git a/mp/src/utils/phonemeextractor/phonemeextractor_ims.vpc b/mp/src/utils/phonemeextractor/phonemeextractor_ims.vpc
index e3df0327..72d7a82b 100644
--- a/mp/src/utils/phonemeextractor/phonemeextractor_ims.vpc
+++ b/mp/src/utils/phonemeextractor/phonemeextractor_ims.vpc
@@ -1,98 +1,98 @@
-//-----------------------------------------------------------------------------

-// PHONEMEEXTRACTOR_IMS.VPC

-//

-//	Project Script

-//-----------------------------------------------------------------------------

-

-$Macro SRCDIR		"..\.."

-$Macro OUTBINDIR	"$SRCDIR\..\game\bin\phonemeextractors"

-

-$Include "$SRCDIR\vpc_scripts\source_dll_base.vpc"

-

-$Configuration

-{

-	$Compiler

-	{

-		$AdditionalIncludeDirectories		"$BASE;../common,../hlfaceposer,../sapi51/include"

-		$PreprocessorDefinitions			"$BASE;PHONEMEEXTRACTOR_EXPORTS"

-	}

-	

-	$Linker

-	{

-		$AdditionalDependencies				"$BASE odbc32.lib odbccp32.lib"

-	}

-}

-

-$Configuration	"Debug"

-{

-	$General

-	{

-		$OutputDirectory			".\Debug_ims"			[$WIN32]

-		$IntermediateDirectory		".\Debug_ims"			[$WIN32]

-	}

-}

-

-$Configuration	"Release"

-{

-	$General

-	{

-		$OutputDirectory			".\Release_ims"			[$WIN32]

-		$IntermediateDirectory		".\Release_ims"			[$WIN32]

-	}

-}

-

-$Project "Phonemeextractor_ims"

-{

-	$Folder	"Source Files"

-	{

-		$File	"extractor_utils.cpp"

-		$File	"$SRCDIR\public\phonemeconverter.cpp"

-		$File	"$SRCDIR\public\sentence.cpp"

-		$File	"phonemeextractor_ims.cpp"

-	}

-

-	$Folder	"Header Files"

-	{

-		$File	"talkback.h"

-	}

-	

-	$Folder	"SAPI Header Files"

-	{

-		$File	"..\sapi51\Include\sapi.h"

-		$File	"..\sapi51\Include\sapiddk.h"

-		$File	"..\sapi51\Include\Spddkhlp.h"

-		$File	"..\sapi51\Include\spdebug.h"

-		$File	"..\sapi51\Include\sperror.h"

-		$File	"..\sapi51\Include\sphelper.h"

-	}

-	

-	$Folder	"Public Header Files"

-	{

-		$File	"$SRCDIR\public\mathlib\amd3dx.h"

-		$File	"$SRCDIR\public\tier0\basetypes.h"

-		$File	"$SRCDIR\public\tier0\commonmacros.h"

-		$File	"$SRCDIR\public\tier0\dbg.h"

-		$File	"$SRCDIR\public\tier0\fasttimer.h"

-		$File	"$SRCDIR\public\appframework\IAppSystem.h"

-		$File	"$SRCDIR\public\mathlib\mathlib.h"

-		$File	"$SRCDIR\public\phonemeconverter.h"

-		$File	"$SRCDIR\public\phonemeextractor\phonemeextractor.h"

-		$File	"$SRCDIR\public\tier0\platform.h"

-		$File	"$SRCDIR\public\tier0\protected_things.h"

-		$File	"$SRCDIR\public\sentence.h"

-		$File	"$SRCDIR\public\string_t.h"

-		$File	"$SRCDIR\public\tier1\strtools.h"

-		$File	"$SRCDIR\public\tier1\utllinkedlist.h"

-		$File	"$SRCDIR\public\tier1\utlmemory.h"

-		$File	"$SRCDIR\public\tier1\utlvector.h"

-		$File	"$SRCDIR\public\mathlib\vector.h"

-		$File	"$SRCDIR\public\mathlib\vector2d.h"

-		$File	"$SRCDIR\public\vstdlib\vstdlib.h"

-	}

-	

-	$Folder	"Link Libraries"

-	{

-		$Lib mathlib

-		$File	"..\sapi51\lib\i386\sapi.lib"

-	}

-}

+//-----------------------------------------------------------------------------
+// PHONEMEEXTRACTOR_IMS.VPC
+//
+//	Project Script
+//-----------------------------------------------------------------------------
+
+$Macro SRCDIR		"..\.."
+$Macro OUTBINDIR	"$SRCDIR\..\game\bin\phonemeextractors"
+
+$Include "$SRCDIR\vpc_scripts\source_dll_base.vpc"
+
+$Configuration
+{
+	$Compiler
+	{
+		$AdditionalIncludeDirectories		"$BASE;../common,../hlfaceposer,../sapi51/include"
+		$PreprocessorDefinitions			"$BASE;PHONEMEEXTRACTOR_EXPORTS"
+	}
+	
+	$Linker
+	{
+		$AdditionalDependencies				"$BASE odbc32.lib odbccp32.lib"
+	}
+}
+
+$Configuration	"Debug"
+{
+	$General
+	{
+		$OutputDirectory			".\Debug_ims"			[$WIN32]
+		$IntermediateDirectory		".\Debug_ims"			[$WIN32]
+	}
+}
+
+$Configuration	"Release"
+{
+	$General
+	{
+		$OutputDirectory			".\Release_ims"			[$WIN32]
+		$IntermediateDirectory		".\Release_ims"			[$WIN32]
+	}
+}
+
+$Project "Phonemeextractor_ims"
+{
+	$Folder	"Source Files"
+	{
+		$File	"extractor_utils.cpp"
+		$File	"$SRCDIR\public\phonemeconverter.cpp"
+		$File	"$SRCDIR\public\sentence.cpp"
+		$File	"phonemeextractor_ims.cpp"
+	}
+
+	$Folder	"Header Files"
+	{
+		$File	"talkback.h"
+	}
+	
+	$Folder	"SAPI Header Files"
+	{
+		$File	"..\sapi51\Include\sapi.h"
+		$File	"..\sapi51\Include\sapiddk.h"
+		$File	"..\sapi51\Include\Spddkhlp.h"
+		$File	"..\sapi51\Include\spdebug.h"
+		$File	"..\sapi51\Include\sperror.h"
+		$File	"..\sapi51\Include\sphelper.h"
+	}
+	
+	$Folder	"Public Header Files"
+	{
+		$File	"$SRCDIR\public\mathlib\amd3dx.h"
+		$File	"$SRCDIR\public\tier0\basetypes.h"
+		$File	"$SRCDIR\public\tier0\commonmacros.h"
+		$File	"$SRCDIR\public\tier0\dbg.h"
+		$File	"$SRCDIR\public\tier0\fasttimer.h"
+		$File	"$SRCDIR\public\appframework\IAppSystem.h"
+		$File	"$SRCDIR\public\mathlib\mathlib.h"
+		$File	"$SRCDIR\public\phonemeconverter.h"
+		$File	"$SRCDIR\public\phonemeextractor\phonemeextractor.h"
+		$File	"$SRCDIR\public\tier0\platform.h"
+		$File	"$SRCDIR\public\tier0\protected_things.h"
+		$File	"$SRCDIR\public\sentence.h"
+		$File	"$SRCDIR\public\string_t.h"
+		$File	"$SRCDIR\public\tier1\strtools.h"
+		$File	"$SRCDIR\public\tier1\utllinkedlist.h"
+		$File	"$SRCDIR\public\tier1\utlmemory.h"
+		$File	"$SRCDIR\public\tier1\utlvector.h"
+		$File	"$SRCDIR\public\mathlib\vector.h"
+		$File	"$SRCDIR\public\mathlib\vector2d.h"
+		$File	"$SRCDIR\public\vstdlib\vstdlib.h"
+	}
+	
+	$Folder	"Link Libraries"
+	{
+		$Lib mathlib
+		$File	"..\sapi51\lib\i386\sapi.lib"
+	}
+}
diff --git a/mp/src/utils/phonemeextractor/talkback.h b/mp/src/utils/phonemeextractor/talkback.h
index 3a1b179a..bb6ee808 100644
--- a/mp/src/utils/phonemeextractor/talkback.h
+++ b/mp/src/utils/phonemeextractor/talkback.h
@@ -1,732 +1,732 @@
-//========= Copyright Valve Corporation, All rights reserved. ============//

-//

-// Purpose: 

-//

-// $NoKeywords: $

-//

-//=============================================================================//

-// =============================================================================

-// Interface to the LIPSinc TalkBack 1.1 library (TalkBack_*.lib).

-//

-// Copyright � 1998-2002 LIPSinc. All rights reserved.

-

-#if !defined(TalkBack_h)

-#define TalkBack_h

-

-#include <stddef.h> // size_t.

-

-// Enforce a C API.

-#if defined(__cplusplus)

-extern "C"

-{

-#endif

-

-// -----------------------------------------------------------------------------

-// Use the preprocessor to make the new API compatible with the old one.

-

-#define TalkbackStartupLibrary                   TalkBackStartupLibrary

-#define TalkbackShutdownLibrary                  TalkBackShutdownLibrary

-#define TalkbackGetVersion                       TalkBackGetVersion

-#define TalkbackGetVersionString                 TalkBackGetVersionString

-#define TalkbackCheckSoundFile                   TalkBackCheckSoundFile

-#define TalkbackCheckSpokenText                  TalkBackCheckSpokenText

-#define TalkbackGetErrorString                   TalkBackGetErrorString

-#define TalkbackGetAnalysis                      TalkBackGetAnalysis

-#define TalkbackFreeAnalysis                     TalkBackFreeAnalysis

-#define TalkbackGetFirstFrameNum                 TalkBackGetFirstFrameNum

-#define TalkbackGetLastFrameNum                  TalkBackGetLastFrameNum

-#define TalkbackGetFrameStartTime                TalkBackGetFrameStartTime

-#define TalkbackGetFrameEndTime                  TalkBackGetFrameEndTime

-#define TalkbackGetNumPhonemes                   TalkBackGetNumPhonemes

-#define TalkbackGetPhonemeEnum                   TalkBackGetPhonemeEnum

-#define TalkbackGetPhonemeStartTime              TalkBackGetPhonemeStartTime

-#define TalkbackGetPhonemeEndTime                TalkBackGetPhonemeEndTime

-#define TalkbackInsertPhoneme                    TalkBackInsertPhoneme

-#define TalkbackDeletePhoneme                    TalkBackDeletePhoneme

-#define TalkbackChangePhonemeStart               TalkBackChangePhonemeStart

-#define TalkbackChangePhonemeEnd                 TalkBackChangePhonemeEnd

-#define TalkbackChangePhonemeEnum                TalkBackChangePhonemeEnum

-#define TalkbackGetNumWords                      TalkBackGetNumWords

-#define TalkbackGetWord                          TalkBackGetWord

-#define TalkbackGetWordStartTime                 TalkBackGetWordStartTime

-#define TalkbackGetWordEndTime                   TalkBackGetWordEndTime

-#define TalkbackGetNumSpeechTargetTracks         TalkBackGetNumSpeechTargetTracks

-#define TalkbackGetNumSpeechTargetKeys           TalkBackGetNumSpeechTargetKeys

-#define TalkbackGetSpeechTargetKeyInfo           TalkBackGetSpeechTargetKeyInfo

-#define TalkbackGetSpeechTargetValueAtFrame      TalkBackGetSpeechTargetValueAtFrame

-#define TalkbackGetDominantSpeechTargetAtFrame   TalkBackGetDominantSpeechTargetAtFrame

-#define TalkbackGetSpeechTargetValueAtTime       TalkBackGetSpeechTargetValueAtTime

-#define TalkbackGetSpeechTargetDerivativesAtTime TalkBackGetSpeechTargetDerivativesAtTime

-#define TalkbackGetNumGestureTracks              TalkBackGetNumGestureTracks

-#define TalkbackGetNumGestureKeys                TalkBackGetNumGestureKeys

-#define TalkbackGetGestureKeyInfo                TalkBackGetGestureKeyInfo

-#define TalkbackGetGestureValueAtFrame           TalkBackGetGestureValueAtFrame

-#define TalkbackGetGestureValueAtTime            TalkBackGetGestureValueAtTime

-#define TalkbackGetGestureDerivativesAtTime      TalkBackGetGestureDerivativesAtTime

-

-// -----------------------------------------------------------------------------

-// For debug builds, set this to a non-zero value to get verbose debugging

-// output from TalkBack.

-

-extern int g_enableTalkBackDebuggingOutput;

-

-// -----------------------------------------------------------------------------

-// Miscellaneous constants.

-

-// For calling TalkBackGetAnalysis() with all defaults.

-#define TALKBACK_DEFAULT_SETTINGS          NULL

-

-// For setting the iSoundText parameter in TalkBackGetAnalysis() to "no text."

-#define TALKBACK_NO_TEXT                   NULL

-

-// Handy constants for TALKBACK_ANALYSIS_SETTINGS fields:

-

-                                           // For setting fSize.

-#define TALKBACK_SETTINGS_SIZE             sizeof(TALKBACK_ANALYSIS_SETTINGS)

-                                           // For setting fFrameRate to the

-                                           // default.

-#define TALKBACK_DEFAULT_FRAME_RATE        30

-                                           // For setting fOptimizeForFlipbook

-                                           // to *not* optimize for flipbook.

-#define TALKBACK_OPTIMIZE_FOR_FLIPBOOK_OFF 0

-                                           // For setting fOptimizeForFlipbook

-                                           // to optimize for flipbook.

-#define TALKBACK_OPTIMIZE_FOR_FLIPBOOK_ON  1

-                                           // For setting fRandomSeed to use the

-                                           // current time to seed the random

-                                           // number generator and thereby get

-                                           // non-deterministic speech gestures.

-#define TALKBACK_RANDOM_SEED               -1

-                                           // For setting fConfigFile to "no

-                                           // config file."

-#define TALKBACK_NO_CONFIG_FILE            NULL

-

-// -----------------------------------------------------------------------------

-// Data types.

-

-// TALKBACK_NOERR if successful, TalkBack error code if not.

-typedef long TALKBACK_ERR;

-

-// Opaque analysis results.

-typedef void TALKBACK_ANALYSIS;

-

-// Speech target.

-typedef long TALKBACK_SPEECH_TARGET;

-

-// Speech gesture.

-typedef long TALKBACK_GESTURE;

-

-// Phoneme.

-typedef long TALKBACK_PHONEME;

-

-// -----------------------------------------------------------------------------

-// Data structures.

-

-#pragma pack(push, 1)

-

-// Optional analysis settings passed to TalkBackGetAnalysis().

-typedef struct

-{

-    // Set this field to sizeof(TALKBACK_ANALYSIS_SETTINGS) before using the

-    // structure.

-    long        fSize;

-    // Frame rate for analysis. This only matters if you will be using *AtFrame

-    // functions.

-    //

-    // Default value: 30 (frames per second).

-    long        fFrameRate;

-    // Set this to 1 to optimize for flipbook output, 0 to do analysis normally.

-    //

-    // Default value: 0 (normal analysis).

-    long        fOptimizeForFlipbook;

-    // Set this to -1 to seed the random number generator with the current time.

-    // Any other number will be used directly for the random number seed, which

-    // is useful if you want repeatable speech gestures. This value does not

-    // influence lip-synching at all.

-    //

-    // Default value: -1 (use current time).

-    long        fRandomSeed;

-    // Path to the configuration (.INI) file with phoneme-to-speech-target

-    // mapping. Set this to NULL to use the default mapping.

-    //

-    // Default value: NULL (use default mapping).

-    char const *fConfigFile;

-} TALKBACK_ANALYSIS_SETTINGS;

-

-typedef struct

-{

-    // Set this field to sizeof(TALKBACK_SOUND_FILE_METRICS) before using the

-    // structure. This will allow the structure to evolve if necessary.

-    size_t     m_size;

-    // Bits per sample.

-    long       m_bitsPerSample;

-    // Sample rate in Hertz.

-    long       m_sampleRate;

-    // Duration of the audio in seconds.

-    double     m_duration;

-    // 1 if the sound file can be analyzed, 0 if not.

-    long       m_canBeAnalyzed;

-    // 1 if the sound file is clipped, 0 if not.

-    long       m_isClipped;

-    // The decibel range of the sound file.

-    double     m_decibelRange;

-    // A quality value for the sound file: the nominal range is 0 to 100. Try

-    // to keep it above 45 for good results.

-    int        m_quality;

-

-    // Added for version 2 of the metrics structure:

-    // ---------------------------------------------

-    // The number of channels in the sound file: 1 for mono, 2 for stereo, etc.

-    long       m_channelCount;

-} TALKBACK_SOUND_FILE_METRICS;

-

-#pragma pack(pop)

-

-// -----------------------------------------------------------------------------

-// Constants.

-

-// TalkBack error codes. Use TalkBackGetErrorString() to return text

-// descriptions for these codes.

-enum

-{

-    // Windows convention: set this bit to indicate an application-defined error

-    // code.

-    BIT29 = (1 << 29),

-    // Success (not an error).

-    TALKBACK_NOERR = 0,

-    // The first error code: useful for iterating through the error codes.

-    TALKBACK_ERROR_FIRST = 4201 | BIT29,

-    // Generic error.

-    TALKBACK_ERROR = TALKBACK_ERROR_FIRST,

-    // TalkBackStartupLibrary() failed [internal error] or was never called.

-    TALKBACK_STARTUP_FAILED_ERR,

-    // TalkBackShutdownLibrary() failed, either because

-    // TalkBackStartupLibrary() was never called or because

-    // TalkBackShutdownLibrary() has already been called.

-    TALKBACK_SHUTDOWN_FAILED_ERR,

-    // The TalkBack data files could not be found [invalid path or missing

-    // files].

-    TALKBACK_CORE_DATA_NOT_FOUND_ERR,

-    // One or more of the parameters are NULL.

-    TALKBACK_NULL_PARAMETER_ERR,

-    // One or more of the parameters is invalid.

-    TALKBACK_INVALID_PARAMETER_ERR,

-    // The analysis object pointer is invalid.

-    TALKBACK_INVALID_ANALYSIS_ERR,

-    // Analysis failed [the sound file cannot be analyzed or an internal error

-    // occurred].

-    TALKBACK_ANALYSIS_FAILED_ERR,

-    // One or more of the indices (track, key, frame, word, phoneme) are

-    // invalid (out of range).

-    TALKBACK_INVALID_INDEX_ERR,

-    // The time parameter is invalid (out of range).

-    TALKBACK_INVALID_TIME_ERR,

-    // A serious internal error occurred in TalkBack; please alert LIPSinc by

-    // sending mail with a description of how the error was triggered to

-    // [email protected].

-    TALKBACK_INTERNAL_ERR,

-    // Could not open the specified sound file.

-    TALKBACK_COULD_NOT_LOAD_SOUND_ERR,

-    // TalkBackStartupLibrary() has not been called.

-    TALKBACK_STARTUP_NOT_CALLED,

-    // The configuration file specified in the TALKBACK_ANALYSIS_SETTINGS

-    // structure is invalid.

-    TALKBACK_CONFIG_PARSE_ERROR,

-    // The last error code: useful for iterating through the error codes.

-    TALKBACK_ERROR_LAST = TALKBACK_CONFIG_PARSE_ERROR

-};

-

-// Default lip-synching track identifiers.

-//

-// NOTE: these track identifiers apply *only* to the default phoneme-to-track

-//       mapping! Consult the TalkBack Reference Guide for more details.

-//

-// NOTE: these values are valid *only* if you use the default mapping and are

-//       provided as a convenience. If you use your own mapping, these values

-//       are invalid and should not be used.

-

-enum

-{

-    TALKBACK_SPEECH_TARGET_INVALID = -1,

-    TALKBACK_SPEECH_TARGET_FIRST = 0,

-    TALKBACK_SPEECH_TARGET_EAT = TALKBACK_SPEECH_TARGET_FIRST, // 0

-    TALKBACK_SPEECH_TARGET_EARTH,                              // 1

-    TALKBACK_SPEECH_TARGET_IF,                                 // 2

-    TALKBACK_SPEECH_TARGET_OX,                                 // 3

-    TALKBACK_SPEECH_TARGET_OAT,                                // 4

-    TALKBACK_SPEECH_TARGET_WET,                                // 5

-    TALKBACK_SPEECH_TARGET_SIZE,                               // 6

-    TALKBACK_SPEECH_TARGET_CHURCH,                             // 7

-    TALKBACK_SPEECH_TARGET_FAVE,                               // 8

-    TALKBACK_SPEECH_TARGET_THOUGH,                             // 9

-    TALKBACK_SPEECH_TARGET_TOLD,                               // 10

-    TALKBACK_SPEECH_TARGET_BUMP,                               // 11

-    TALKBACK_SPEECH_TARGET_NEW,                                // 12

-    TALKBACK_SPEECH_TARGET_ROAR,                               // 13

-    TALKBACK_SPEECH_TARGET_CAGE,                               // 14

-    TALKBACK_SPEECH_TARGET_LAST = TALKBACK_SPEECH_TARGET_CAGE, // 14

-    TALKBACK_NUM_SPEECH_TARGETS                                // 15 (0..14)

-};

-

-// Speech gesture track identifiers.

-

-enum

-{

-    TALKBACK_GESTURE_INVALID = -1,

-    TALKBACK_GESTURE_FIRST = 0,

-    TALKBACK_GESTURE_EYEBROW_RAISE_LEFT = TALKBACK_GESTURE_FIRST, // 0

-    TALKBACK_GESTURE_EYEBROW_RAISE_RIGHT,                         // 1

-    TALKBACK_GESTURE_BLINK_LEFT,                                  // 2

-    TALKBACK_GESTURE_BLINK_RIGHT,                                 // 3

-    TALKBACK_GESTURE_HEAD_BEND,                                   // 4

-    TALKBACK_GESTURE_HEAD_SIDE_SIDE,                              // 5

-    TALKBACK_GESTURE_HEAD_TWIST,                                  // 6

-    TALKBACK_GESTURE_EYE_SIDE_SIDE_LEFT,                          // 7

-    TALKBACK_GESTURE_EYE_SIDE_SIDE_RIGHT,                         // 8

-    TALKBACK_GESTURE_EYE_UP_DOWN_LEFT,                            // 9

-    TALKBACK_GESTURE_EYE_UP_DOWN_RIGHT,                           // 10

-    TALKBACK_GESTURE_LAST = TALKBACK_GESTURE_EYE_UP_DOWN_RIGHT,   // 10

-    TALKBACK_NUM_GESTURES                                         // 11 (0..10)

-};

-

-// Phoneme identifiers.

-

-enum

-{

-    TALKBACK_PHONEME_INVALID = -1,

-    TALKBACK_PHONEME_FIRST = 0,

-    TALKBACK_PHONEME_IY = TALKBACK_PHONEME_FIRST, // 0

-    TALKBACK_PHONEME_IH,                          // 1

-    TALKBACK_PHONEME_EH,                          // 2

-    TALKBACK_PHONEME_EY,                          // 3

-    TALKBACK_PHONEME_AE,                          // 4

-    TALKBACK_PHONEME_AA,                          // 5

-    TALKBACK_PHONEME_AW,                          // 6

-    TALKBACK_PHONEME_AY,                          // 7

-    TALKBACK_PHONEME_AH,                          // 8

-    TALKBACK_PHONEME_AO,                          // 9

-    TALKBACK_PHONEME_OY,                          // 10

-    TALKBACK_PHONEME_OW,                          // 11

-    TALKBACK_PHONEME_UH,                          // 12

-    TALKBACK_PHONEME_UW,                          // 13

-    TALKBACK_PHONEME_ER,                          // 14

-    TALKBACK_PHONEME_AX,                          // 15

-    TALKBACK_PHONEME_S,                           // 16

-    TALKBACK_PHONEME_SH,                          // 17

-    TALKBACK_PHONEME_Z,                           // 18

-    TALKBACK_PHONEME_ZH,                          // 19

-    TALKBACK_PHONEME_F,                           // 20

-    TALKBACK_PHONEME_TH,                          // 21

-    TALKBACK_PHONEME_V,                           // 22

-    TALKBACK_PHONEME_DH,                          // 23

-    TALKBACK_PHONEME_M,                           // 24

-    TALKBACK_PHONEME_N,                           // 25

-    TALKBACK_PHONEME_NG,                          // 26

-    TALKBACK_PHONEME_L,                           // 27

-    TALKBACK_PHONEME_R,                           // 28

-    TALKBACK_PHONEME_W,                           // 29

-    TALKBACK_PHONEME_Y,                           // 30

-    TALKBACK_PHONEME_HH,                          // 31

-    TALKBACK_PHONEME_B,                           // 32

-    TALKBACK_PHONEME_D,                           // 33

-    TALKBACK_PHONEME_JH,                          // 34

-    TALKBACK_PHONEME_G,                           // 35

-    TALKBACK_PHONEME_P,                           // 36

-    TALKBACK_PHONEME_T,                           // 37

-    TALKBACK_PHONEME_K,                           // 38

-    TALKBACK_PHONEME_CH,                          // 39

-    TALKBACK_PHONEME_SIL,                         // 40

-    TALKBACK_PHONEME_LAST = TALKBACK_PHONEME_SIL, // 40

-    TALKBACK_NUM_PHONEMES                         // 41 (0..40)

-};

-

-// -----------------------------------------------------------------------------

-// Function declarations.

-

-// ---------------------------

-// Startup/shutdown functions.

-// ---------------------------

-

-// Must be the first function called when using TalkBack.

-TALKBACK_ERR                   // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackStartupLibrary(

-    char const *iCoreDataDir); // IN:      full path of folder containing TalkBack data files.

-

-// Should be the last function called when using TalkBack.

-TALKBACK_ERR               // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackShutdownLibrary(); // IN:      nothing.

-

-// ------------------

-// Version functions.

-// ------------------

-

-// Gets the TalkBack version number.

-TALKBACK_ERR          // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetVersion(

-    long *oMajor,     // OUT:     major version number.

-    long *oMinor,     // OUT:     minor version number.

-    long *oRevision); // OUT:     revision version number.

-

-// Gets the TalkBack version number as a string.

-TALKBACK_ERR         // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetVersionString(

-    long  iMaxChars, // IN:      size of version string buffer.

-    char *oVersion); // OUT:     version string buffer.

-

-// ------------------

-// Utility functions.

-// ------------------

-

-// Checks whether a sound file can be analyzed and returns some quality metrics.

-//

-// NOTE: this function is deprecated and has been supplanted by

-// TalkBackGetSoundFileMetrics().

-TALKBACK_ERR                    // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackCheckSoundFile(

-    char const *iSoundFileName, // IN:      name of sound file to be checked.

-    long       *oCanBeAnalyzed, // OUT:     1 if sound can be analyzed, 0 if not.

-    long       *oIsClipped,     // OUT:     1 if sound is clipped, 0 if not.

-    double     *oDecibelRange); // OUT:     used decibel range of sound.

-

-// Returns metrics for the specified sound file.

-TALKBACK_ERR                                     // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetSoundFileMetrics(

-    char const                  *iSoundFileName, // IN:      name of sound file to be checked.

-    TALKBACK_SOUND_FILE_METRICS *ioMetrics);     // IN/OUT:  address of a structure where the metrics will be stored.

-

-// Checks whether text can be used for text-based analysis, returning the text

-// as it will be analyzed.

-TALKBACK_ERR                    // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackCheckSpokenText(

-    char const *iSpokenText,    // IN:      text to check.

-    long        iMaxChars,      // IN:      size of analyzed text buffer.

-    char       *oAnalyzedText); // OUT:     buffer for text as it will be analyzed.

-

-// Convert a TalkBack error code to a description string.

-TALKBACK_ERR                    // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetErrorString(

-    TALKBACK_ERR iErrorCode,    // IN:      TalkBack error code to convert.

-    long         iMaxChars,     // IN:      size of the buffer.

-    char        *oErrorString); // OUT:     buffer for the description string.

-

-// Gets the error code and text for the most recent TalkBack error.

-TALKBACK_ERR                     // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetLastError(

-    long          iMaxChars,    // IN:      size of the buffer.

-    char         *oErrorString, // OUT:     buffer for the description string.

-    TALKBACK_ERR *oErrorCode);  // OUT:     most recent TalkBack error code.

-

-// -------------------

-// Analysis functions.

-// -------------------

-

-// Gets an opaque TALKBACK_ANALYSIS object. This object is then queried with the

-// TalkBackGet* functions below.

-TALKBACK_ERR                                    // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetAnalysis(

-    TALKBACK_ANALYSIS         **ioAnalysis,     // IN/OUT:  address of a TALKBACK_ANALYSIS *variable where analysis will be stored.

-    char const                 *iSoundFileName, // IN:      name of the sound file to analyze.

-    char const                 *iSoundText,     // IN:      text spoken in sound file (can be NULL to use textless analysis).

-    TALKBACK_ANALYSIS_SETTINGS *iSettings);     // IN:      pointer to a TALKBACK_ANALYSIS_SETTINGS structure (can be NULL for defaults).

-

-// Frees an opaque TALKBACK_ANALYSIS object. This releases all memory used by

-// the analysis.

-TALKBACK_ERR                         // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackFreeAnalysis(

-    TALKBACK_ANALYSIS **ioAnalysis); // IN/OUT:  analysis to free.

-

-// #######################################################################

-// NOTE: all functions from this point on require a valid analysis object.

-// #######################################################################

-

-// ------------------------

-// Speech target functions.

-// ------------------------

-

-// Gets the number of speech target tracks.

-TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetNumSpeechTargetTracks(

-    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long              *oResult);  // OUT:     number of speech target tracks.

-

-// Gets the number of keys in the specified speech target track.

-TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetNumSpeechTargetKeys(

-    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long               iTrackNum, // IN:      speech target track.

-    long              *oResult);  // OUT:     number of keys in the speech target track.

-

-// Gets key information (time, value, derivative in, and derivative out) for the

-// specified key in the specified speech target track.

-TALKBACK_ERR                            // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetSpeechTargetKeyInfo(

-    TALKBACK_ANALYSIS *iAnalysis,       // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long               iTrackNum,       // IN:      speech target track.

-    long               iKeyNum,         // IN:      speech target key.

-    double            *oTime,           // OUT:     time of key.

-    double            *oValue,          // OUT:     value of key.

-    double            *oDerivativeIn,   // OUT:     incoming derivative of key.

-    double            *oDerivativeOut); // OUT:     outgoing derivative of key.

-

-// Gets the value of the function curve for the specified speech target track at

-// the specified time.

-TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetSpeechTargetValueAtTime(

-    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long               iTrackNum, // IN:      speech target track.

-    double             iTime,     // IN:      time in seconds.

-    double            *oResult);  // OUT:     value of the function curve.

-

-// Gets the derivatives of the function curve for the specified speech target

-// track at the specified time.

-TALKBACK_ERR                            // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetSpeechTargetDerivativesAtTime(

-    TALKBACK_ANALYSIS *iAnalysis,       // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long               iTrackNum,       // IN:      speech target track.

-    double             iTime,           // IN:      time in seconds.

-    double            *oDerivativeIn,   // OUT:     value of the incoming derivative of the function curve.

-    double            *oDerivativeOut); // OUT:     value of the outgoing derivative of the function curve.

-

-// -------------------------

-// Speech gesture functions.

-// -------------------------

-

-// Gets the number of speech gesture tracks.

-TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetNumGestureTracks(

-    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long              *oResult);  // OUT:     number of speech gesture tracks

-

-// Gets the number of keys in the specified speech gesture track.

-TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetNumGestureKeys(

-    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long               iTrackNum, // IN:      speech gesture track.

-    long              *oResult);  // OUT:     number of keys in the speech gesture track.

-

-// Gets key information (time, value, derivative in, and derivative out) for the

-// specified key in the specified speech gesture track.

-TALKBACK_ERR                            // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetGestureKeyInfo(

-    TALKBACK_ANALYSIS *iAnalysis,       // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long               iTrackNum,       // IN:      speech gesture track.

-    long               iKeyNum,         // IN:      speech gesture key.

-    double            *oTime,           // OUT:     time of key.

-    double            *oValue,          // OUT:     value of key.

-    double            *oDerivativeIn,   // OUT:     incoming derivative of key.

-    double            *oDerivativeOut); // OUT:     outgoing derivative of key.

-

-// Gets the value of the function curve for the specified speech gesture track

-// at the specified time.

-TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetGestureValueAtTime(

-    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long               iTrackNum, // IN:      speech gesture track.

-    double             iTime,     // IN:      time in seconds.

-    double            *oResult);  // OUT:     value of the function curve.

-

-// Gets the derivatives of the function curve for the specified speech gesture

-// track at the specified time.

-TALKBACK_ERR                            // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetGestureDerivativesAtTime(

-    TALKBACK_ANALYSIS *iAnalysis,       // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long               iTrackNum,       // IN:      speech gesture track.

-    double             iTime,           // IN:      time in seconds.

-    double            *oDerivativeIn,   // OUT:     value of the incoming derivative of the function curve.

-    double            *oDerivativeOut); // OUT:     value of the outgoing derivative of the function curve.

-

-// ----------------

-// Frame functions.

-// ----------------

-

-// NOTE: these functions use the frame rate specified in the

-// TALKBACK_ANALYSIS_SETTINGS structure passed to TalkBackGetAnalysis() and

-// default to 30 fps (TALKBACK_DEFAULT_FRAME_RATE) if the structure pointer was

-// NULL.

-

-// Gets the first frame number.

-TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetFirstFrameNum(

-    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long              *oResult);  // OUT:     number of the first frame.

-

-// Gets the last frame number.

-TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetLastFrameNum(

-    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long              *oResult);  // OUT:     number of the last frame.

-

-// Gets the start time of the specified frame.

-TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetFrameStartTime(

-    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long               iFrameNum, // IN:      frame.

-    double            *oResult);  // OUT:     start time of the frame in seconds.

-

-// Gets the end time of the specified frame.

-TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetFrameEndTime(

-    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long               iFrameNum, // IN:      frame.

-    double            *oResult);  // OUT:     end time of the frame in seconds.

-

-// Gets the value of the function curve for a speech target integrated over the

-// specified frame.

-TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetSpeechTargetValueAtFrame(

-    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long               iTrackNum, // IN:      speech target track.

-    long               iFrameNum, // IN:      frame number.

-    double            *oResult);  // OUT:     value of the function curve integrated over the frame.

-

-// Gets the dominant speech target at the specified frame.

-//

-// NOTE: this function is meant to be used in flipbook mode only.

-TALKBACK_ERR                                // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetDominantSpeechTargetAtFrame(

-    TALKBACK_ANALYSIS      *iAnalysis,      // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long                    iFrameNum,      // IN:      frame number.

-    TALKBACK_SPEECH_TARGET *oSpeechTarget); // OUT:     dominant speech target.

-

-// Gets the value of the function curve for a speech gesture integrated over the

-// specified frame.

-TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetGestureValueAtFrame(

-    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long               iTrackNum, // IN:      speech gesture track.

-    long               iFrameNum, // IN:      frame number.

-    double            *oResult);  // OUT:     value of the function curve integrated over the frame.

-

-// ------------------

-// Phoneme functions.

-// ------------------

-

-// Gets the number of phonemes.

-TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetNumPhonemes(

-    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long              *oResult);  // OUT:     number of phonemes.

-

-// Gets the enumeration of the specified phoneme.

-TALKBACK_ERR                        // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetPhonemeEnum(

-    TALKBACK_ANALYSIS *iAnalysis,   // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long               iPhonemeNum, // IN:      phoneme.

-    TALKBACK_PHONEME  *oResult);    // OUT:     enumeration of the specified phoneme.

-

-// Gets the start time of the specified phoneme.

-TALKBACK_ERR                        // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetPhonemeStartTime(

-    TALKBACK_ANALYSIS *iAnalysis,   // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long               iPhonemeNum, // IN:      phoneme.

-    double            *oResult);    // OUT:     start time of the phoneme in seconds.

-

-// Gets the end time of the specified phoneme.

-TALKBACK_ERR                        // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetPhonemeEndTime(

-    TALKBACK_ANALYSIS *iAnalysis,   // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long               iPhonemeNum, // IN:      phoneme.

-    double            *oResult);    // OUT:     end time of the phoneme in seconds.

-

-// ---------------

-// Word functions.

-// ---------------

-

-// NOTE: these functions only yield data for text-based analysis.

-

-// Gets the number of words.

-TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetNumWords(

-    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long              *oResult);  // OUT:     number of words.

-

-// Gets the text of the specified word.

-TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetWord(

-    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long               iWordNum,  // IN:      word.

-    long               iMaxChars, // IN:      size of word buffer.

-    char              *oWord);    // OUT:     word buffer.

-

-// Gets the start time of the specified word.

-TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetWordStartTime(

-    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long               iWordNum,  // IN:      word.

-    double            *oResult);  // OUT:     start time of the word in seconds.

-

-// Gets the end time of the specified word.

-TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackGetWordEndTime(

-    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long               iWordNum,  // IN:      word.

-    double            *oResult);  // OUT:     end time of the word in seconds.

-

-// --------------------------

-// Phoneme editing functions.

-// --------------------------

-

-// Use these functions to modify the phoneme list after you get an opaque

-// analysis object from TalkBackGetAnalysis(). After modifying the phoneme list

-// in the opaque analysis object, subsequent TalkBackGet* calls on that opaque

-// analysis object for speech target (lip-synching) data will return values

-// based on the modified phoneme list. However, speech gesture data is not

-// affected by phoneme editing.

-//

-// NOTE: phoneme editing is only provided in order to support Ventriloquist-like

-// applications where tweaking of the phoneme segmenation (and subsequent

-// recalculation of the animation data) is required. Most customers probably

-// won't need this functionality.

-

-// Inserts a phoneme at the specified position in the specified manner.

-TALKBACK_ERR                            // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackInsertPhoneme(

-    TALKBACK_ANALYSIS *iAnalysis,       // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    TALKBACK_PHONEME   iPhoneme,        // IN:      enumeration of phoneme to insert.

-    long               iInsertPosition, // IN:      position (phoneme number) at which to insert.

-    int                iInsertBefore);  // IN:      manner of insertion:

-                                        //            0 means put phoneme after insert position;

-                                        //            1 means put phoneme before insert position.

-

-// Deletes the specified phoneme.

-TALKBACK_ERR                              // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackDeletePhoneme(

-    TALKBACK_ANALYSIS *iAnalysis,         // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long               iPhonemeToDelete); // IN:      phoneme to delete.

-

-// Changes the start time of the specified phoneme.

-//

-// NOTE: the start time specified may not be the actual start time for a number

-// of reasons, most notably if the specified start time will make the phoneme

-// too short. This function returns the actual start time so the caller can

-// check the result without having to query the phoneme.

-TALKBACK_ERR                             // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackChangePhonemeStart(

-    TALKBACK_ANALYSIS *iAnalysis,        // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long               iPhonemeToChange, // IN:      phoneme to change.

-    double            *ioNewTime);       // IN/OUT:  new start time value in seconds (in); actual start time (out).

-

-// Changes the end time of the specified phoneme.

-//

-// NOTE: the end time specified may not be the actual end time for a number of

-// reasons, most notably if the specified end time will make the phoneme too

-// short. This function returns the actual end time so the caller can check the

-// result without having to query the phoneme.

-TALKBACK_ERR                             // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackChangePhonemeEnd(

-    TALKBACK_ANALYSIS *iAnalysis,        // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long               iPhonemeToChange, // IN:      phoneme to change.

-    double            *ioNewTime);       // IN/OUT:  new end time value in seconds (in); actual end time (out).

-

-// Changes the enumeration of the specified phoneme.

-TALKBACK_ERR                             // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.

-TalkBackChangePhonemeEnum(

-    TALKBACK_ANALYSIS *iAnalysis,        // IN:      opaque analysis object returned by TalkBackGetAnalysis().

-    long               iPhonemeToChange, // IN:      phoneme to change.

-    TALKBACK_PHONEME   iNewPhoneme);     // IN:      new phoneme enumeration.

-

-#if defined(__cplusplus)

-}

-#endif

-

-#endif

+//========= Copyright Valve Corporation, All rights reserved. ============//
+//
+// Purpose: 
+//
+// $NoKeywords: $
+//
+//=============================================================================//
+// =============================================================================
+// Interface to the LIPSinc TalkBack 1.1 library (TalkBack_*.lib).
+//
+// Copyright � 1998-2002 LIPSinc. All rights reserved.
+
+#if !defined(TalkBack_h)
+#define TalkBack_h
+
+#include <stddef.h> // size_t.
+
+// Enforce a C API.
+#if defined(__cplusplus)
+extern "C"
+{
+#endif
+
+// -----------------------------------------------------------------------------
+// Use the preprocessor to make the new API compatible with the old one.
+
+#define TalkbackStartupLibrary                   TalkBackStartupLibrary
+#define TalkbackShutdownLibrary                  TalkBackShutdownLibrary
+#define TalkbackGetVersion                       TalkBackGetVersion
+#define TalkbackGetVersionString                 TalkBackGetVersionString
+#define TalkbackCheckSoundFile                   TalkBackCheckSoundFile
+#define TalkbackCheckSpokenText                  TalkBackCheckSpokenText
+#define TalkbackGetErrorString                   TalkBackGetErrorString
+#define TalkbackGetAnalysis                      TalkBackGetAnalysis
+#define TalkbackFreeAnalysis                     TalkBackFreeAnalysis
+#define TalkbackGetFirstFrameNum                 TalkBackGetFirstFrameNum
+#define TalkbackGetLastFrameNum                  TalkBackGetLastFrameNum
+#define TalkbackGetFrameStartTime                TalkBackGetFrameStartTime
+#define TalkbackGetFrameEndTime                  TalkBackGetFrameEndTime
+#define TalkbackGetNumPhonemes                   TalkBackGetNumPhonemes
+#define TalkbackGetPhonemeEnum                   TalkBackGetPhonemeEnum
+#define TalkbackGetPhonemeStartTime              TalkBackGetPhonemeStartTime
+#define TalkbackGetPhonemeEndTime                TalkBackGetPhonemeEndTime
+#define TalkbackInsertPhoneme                    TalkBackInsertPhoneme
+#define TalkbackDeletePhoneme                    TalkBackDeletePhoneme
+#define TalkbackChangePhonemeStart               TalkBackChangePhonemeStart
+#define TalkbackChangePhonemeEnd                 TalkBackChangePhonemeEnd
+#define TalkbackChangePhonemeEnum                TalkBackChangePhonemeEnum
+#define TalkbackGetNumWords                      TalkBackGetNumWords
+#define TalkbackGetWord                          TalkBackGetWord
+#define TalkbackGetWordStartTime                 TalkBackGetWordStartTime
+#define TalkbackGetWordEndTime                   TalkBackGetWordEndTime
+#define TalkbackGetNumSpeechTargetTracks         TalkBackGetNumSpeechTargetTracks
+#define TalkbackGetNumSpeechTargetKeys           TalkBackGetNumSpeechTargetKeys
+#define TalkbackGetSpeechTargetKeyInfo           TalkBackGetSpeechTargetKeyInfo
+#define TalkbackGetSpeechTargetValueAtFrame      TalkBackGetSpeechTargetValueAtFrame
+#define TalkbackGetDominantSpeechTargetAtFrame   TalkBackGetDominantSpeechTargetAtFrame
+#define TalkbackGetSpeechTargetValueAtTime       TalkBackGetSpeechTargetValueAtTime
+#define TalkbackGetSpeechTargetDerivativesAtTime TalkBackGetSpeechTargetDerivativesAtTime
+#define TalkbackGetNumGestureTracks              TalkBackGetNumGestureTracks
+#define TalkbackGetNumGestureKeys                TalkBackGetNumGestureKeys
+#define TalkbackGetGestureKeyInfo                TalkBackGetGestureKeyInfo
+#define TalkbackGetGestureValueAtFrame           TalkBackGetGestureValueAtFrame
+#define TalkbackGetGestureValueAtTime            TalkBackGetGestureValueAtTime
+#define TalkbackGetGestureDerivativesAtTime      TalkBackGetGestureDerivativesAtTime
+
+// -----------------------------------------------------------------------------
+// For debug builds, set this to a non-zero value to get verbose debugging
+// output from TalkBack.
+
+extern int g_enableTalkBackDebuggingOutput;
+
+// -----------------------------------------------------------------------------
+// Miscellaneous constants.
+
+// For calling TalkBackGetAnalysis() with all defaults.
+#define TALKBACK_DEFAULT_SETTINGS          NULL
+
+// For setting the iSoundText parameter in TalkBackGetAnalysis() to "no text."
+#define TALKBACK_NO_TEXT                   NULL
+
+// Handy constants for TALKBACK_ANALYSIS_SETTINGS fields:
+
+                                           // For setting fSize.
+#define TALKBACK_SETTINGS_SIZE             sizeof(TALKBACK_ANALYSIS_SETTINGS)
+                                           // For setting fFrameRate to the
+                                           // default.
+#define TALKBACK_DEFAULT_FRAME_RATE        30
+                                           // For setting fOptimizeForFlipbook
+                                           // to *not* optimize for flipbook.
+#define TALKBACK_OPTIMIZE_FOR_FLIPBOOK_OFF 0
+                                           // For setting fOptimizeForFlipbook
+                                           // to optimize for flipbook.
+#define TALKBACK_OPTIMIZE_FOR_FLIPBOOK_ON  1
+                                           // For setting fRandomSeed to use the
+                                           // current time to seed the random
+                                           // number generator and thereby get
+                                           // non-deterministic speech gestures.
+#define TALKBACK_RANDOM_SEED               -1
+                                           // For setting fConfigFile to "no
+                                           // config file."
+#define TALKBACK_NO_CONFIG_FILE            NULL
+
+// -----------------------------------------------------------------------------
+// Data types.
+
+// TALKBACK_NOERR if successful, TalkBack error code if not.
+typedef long TALKBACK_ERR;
+
+// Opaque analysis results.
+typedef void TALKBACK_ANALYSIS;
+
+// Speech target.
+typedef long TALKBACK_SPEECH_TARGET;
+
+// Speech gesture.
+typedef long TALKBACK_GESTURE;
+
+// Phoneme.
+typedef long TALKBACK_PHONEME;
+
+// -----------------------------------------------------------------------------
+// Data structures.
+
+#pragma pack(push, 1)
+
+// Optional analysis settings passed to TalkBackGetAnalysis().
+typedef struct
+{
+    // Set this field to sizeof(TALKBACK_ANALYSIS_SETTINGS) before using the
+    // structure.
+    long        fSize;
+    // Frame rate for analysis. This only matters if you will be using *AtFrame
+    // functions.
+    //
+    // Default value: 30 (frames per second).
+    long        fFrameRate;
+    // Set this to 1 to optimize for flipbook output, 0 to do analysis normally.
+    //
+    // Default value: 0 (normal analysis).
+    long        fOptimizeForFlipbook;
+    // Set this to -1 to seed the random number generator with the current time.
+    // Any other number will be used directly for the random number seed, which
+    // is useful if you want repeatable speech gestures. This value does not
+    // influence lip-synching at all.
+    //
+    // Default value: -1 (use current time).
+    long        fRandomSeed;
+    // Path to the configuration (.INI) file with phoneme-to-speech-target
+    // mapping. Set this to NULL to use the default mapping.
+    //
+    // Default value: NULL (use default mapping).
+    char const *fConfigFile;
+} TALKBACK_ANALYSIS_SETTINGS;
+
+typedef struct
+{
+    // Set this field to sizeof(TALKBACK_SOUND_FILE_METRICS) before using the
+    // structure. This will allow the structure to evolve if necessary.
+    size_t     m_size;
+    // Bits per sample.
+    long       m_bitsPerSample;
+    // Sample rate in Hertz.
+    long       m_sampleRate;
+    // Duration of the audio in seconds.
+    double     m_duration;
+    // 1 if the sound file can be analyzed, 0 if not.
+    long       m_canBeAnalyzed;
+    // 1 if the sound file is clipped, 0 if not.
+    long       m_isClipped;
+    // The decibel range of the sound file.
+    double     m_decibelRange;
+    // A quality value for the sound file: the nominal range is 0 to 100. Try
+    // to keep it above 45 for good results.
+    int        m_quality;
+
+    // Added for version 2 of the metrics structure:
+    // ---------------------------------------------
+    // The number of channels in the sound file: 1 for mono, 2 for stereo, etc.
+    long       m_channelCount;
+} TALKBACK_SOUND_FILE_METRICS;
+
+#pragma pack(pop)
+
+// -----------------------------------------------------------------------------
+// Constants.
+
+// TalkBack error codes. Use TalkBackGetErrorString() to return text
+// descriptions for these codes.
+enum
+{
+    // Windows convention: set this bit to indicate an application-defined error
+    // code.
+    BIT29 = (1 << 29),
+    // Success (not an error).
+    TALKBACK_NOERR = 0,
+    // The first error code: useful for iterating through the error codes.
+    TALKBACK_ERROR_FIRST = 4201 | BIT29,
+    // Generic error.
+    TALKBACK_ERROR = TALKBACK_ERROR_FIRST,
+    // TalkBackStartupLibrary() failed [internal error] or was never called.
+    TALKBACK_STARTUP_FAILED_ERR,
+    // TalkBackShutdownLibrary() failed, either because
+    // TalkBackStartupLibrary() was never called or because
+    // TalkBackShutdownLibrary() has already been called.
+    TALKBACK_SHUTDOWN_FAILED_ERR,
+    // The TalkBack data files could not be found [invalid path or missing
+    // files].
+    TALKBACK_CORE_DATA_NOT_FOUND_ERR,
+    // One or more of the parameters are NULL.
+    TALKBACK_NULL_PARAMETER_ERR,
+    // One or more of the parameters is invalid.
+    TALKBACK_INVALID_PARAMETER_ERR,
+    // The analysis object pointer is invalid.
+    TALKBACK_INVALID_ANALYSIS_ERR,
+    // Analysis failed [the sound file cannot be analyzed or an internal error
+    // occurred].
+    TALKBACK_ANALYSIS_FAILED_ERR,
+    // One or more of the indices (track, key, frame, word, phoneme) are
+    // invalid (out of range).
+    TALKBACK_INVALID_INDEX_ERR,
+    // The time parameter is invalid (out of range).
+    TALKBACK_INVALID_TIME_ERR,
+    // A serious internal error occurred in TalkBack; please alert LIPSinc by
+    // sending mail with a description of how the error was triggered to
+    // [email protected].
+    TALKBACK_INTERNAL_ERR,
+    // Could not open the specified sound file.
+    TALKBACK_COULD_NOT_LOAD_SOUND_ERR,
+    // TalkBackStartupLibrary() has not been called.
+    TALKBACK_STARTUP_NOT_CALLED,
+    // The configuration file specified in the TALKBACK_ANALYSIS_SETTINGS
+    // structure is invalid.
+    TALKBACK_CONFIG_PARSE_ERROR,
+    // The last error code: useful for iterating through the error codes.
+    TALKBACK_ERROR_LAST = TALKBACK_CONFIG_PARSE_ERROR
+};
+
+// Default lip-synching track identifiers.
+//
+// NOTE: these track identifiers apply *only* to the default phoneme-to-track
+//       mapping! Consult the TalkBack Reference Guide for more details.
+//
+// NOTE: these values are valid *only* if you use the default mapping and are
+//       provided as a convenience. If you use your own mapping, these values
+//       are invalid and should not be used.
+
+enum
+{
+    TALKBACK_SPEECH_TARGET_INVALID = -1,
+    TALKBACK_SPEECH_TARGET_FIRST = 0,
+    TALKBACK_SPEECH_TARGET_EAT = TALKBACK_SPEECH_TARGET_FIRST, // 0
+    TALKBACK_SPEECH_TARGET_EARTH,                              // 1
+    TALKBACK_SPEECH_TARGET_IF,                                 // 2
+    TALKBACK_SPEECH_TARGET_OX,                                 // 3
+    TALKBACK_SPEECH_TARGET_OAT,                                // 4
+    TALKBACK_SPEECH_TARGET_WET,                                // 5
+    TALKBACK_SPEECH_TARGET_SIZE,                               // 6
+    TALKBACK_SPEECH_TARGET_CHURCH,                             // 7
+    TALKBACK_SPEECH_TARGET_FAVE,                               // 8
+    TALKBACK_SPEECH_TARGET_THOUGH,                             // 9
+    TALKBACK_SPEECH_TARGET_TOLD,                               // 10
+    TALKBACK_SPEECH_TARGET_BUMP,                               // 11
+    TALKBACK_SPEECH_TARGET_NEW,                                // 12
+    TALKBACK_SPEECH_TARGET_ROAR,                               // 13
+    TALKBACK_SPEECH_TARGET_CAGE,                               // 14
+    TALKBACK_SPEECH_TARGET_LAST = TALKBACK_SPEECH_TARGET_CAGE, // 14
+    TALKBACK_NUM_SPEECH_TARGETS                                // 15 (0..14)
+};
+
+// Speech gesture track identifiers.
+
+enum
+{
+    TALKBACK_GESTURE_INVALID = -1,
+    TALKBACK_GESTURE_FIRST = 0,
+    TALKBACK_GESTURE_EYEBROW_RAISE_LEFT = TALKBACK_GESTURE_FIRST, // 0
+    TALKBACK_GESTURE_EYEBROW_RAISE_RIGHT,                         // 1
+    TALKBACK_GESTURE_BLINK_LEFT,                                  // 2
+    TALKBACK_GESTURE_BLINK_RIGHT,                                 // 3
+    TALKBACK_GESTURE_HEAD_BEND,                                   // 4
+    TALKBACK_GESTURE_HEAD_SIDE_SIDE,                              // 5
+    TALKBACK_GESTURE_HEAD_TWIST,                                  // 6
+    TALKBACK_GESTURE_EYE_SIDE_SIDE_LEFT,                          // 7
+    TALKBACK_GESTURE_EYE_SIDE_SIDE_RIGHT,                         // 8
+    TALKBACK_GESTURE_EYE_UP_DOWN_LEFT,                            // 9
+    TALKBACK_GESTURE_EYE_UP_DOWN_RIGHT,                           // 10
+    TALKBACK_GESTURE_LAST = TALKBACK_GESTURE_EYE_UP_DOWN_RIGHT,   // 10
+    TALKBACK_NUM_GESTURES                                         // 11 (0..10)
+};
+
+// Phoneme identifiers.
+
+enum
+{
+    TALKBACK_PHONEME_INVALID = -1,
+    TALKBACK_PHONEME_FIRST = 0,
+    TALKBACK_PHONEME_IY = TALKBACK_PHONEME_FIRST, // 0
+    TALKBACK_PHONEME_IH,                          // 1
+    TALKBACK_PHONEME_EH,                          // 2
+    TALKBACK_PHONEME_EY,                          // 3
+    TALKBACK_PHONEME_AE,                          // 4
+    TALKBACK_PHONEME_AA,                          // 5
+    TALKBACK_PHONEME_AW,                          // 6
+    TALKBACK_PHONEME_AY,                          // 7
+    TALKBACK_PHONEME_AH,                          // 8
+    TALKBACK_PHONEME_AO,                          // 9
+    TALKBACK_PHONEME_OY,                          // 10
+    TALKBACK_PHONEME_OW,                          // 11
+    TALKBACK_PHONEME_UH,                          // 12
+    TALKBACK_PHONEME_UW,                          // 13
+    TALKBACK_PHONEME_ER,                          // 14
+    TALKBACK_PHONEME_AX,                          // 15
+    TALKBACK_PHONEME_S,                           // 16
+    TALKBACK_PHONEME_SH,                          // 17
+    TALKBACK_PHONEME_Z,                           // 18
+    TALKBACK_PHONEME_ZH,                          // 19
+    TALKBACK_PHONEME_F,                           // 20
+    TALKBACK_PHONEME_TH,                          // 21
+    TALKBACK_PHONEME_V,                           // 22
+    TALKBACK_PHONEME_DH,                          // 23
+    TALKBACK_PHONEME_M,                           // 24
+    TALKBACK_PHONEME_N,                           // 25
+    TALKBACK_PHONEME_NG,                          // 26
+    TALKBACK_PHONEME_L,                           // 27
+    TALKBACK_PHONEME_R,                           // 28
+    TALKBACK_PHONEME_W,                           // 29
+    TALKBACK_PHONEME_Y,                           // 30
+    TALKBACK_PHONEME_HH,                          // 31
+    TALKBACK_PHONEME_B,                           // 32
+    TALKBACK_PHONEME_D,                           // 33
+    TALKBACK_PHONEME_JH,                          // 34
+    TALKBACK_PHONEME_G,                           // 35
+    TALKBACK_PHONEME_P,                           // 36
+    TALKBACK_PHONEME_T,                           // 37
+    TALKBACK_PHONEME_K,                           // 38
+    TALKBACK_PHONEME_CH,                          // 39
+    TALKBACK_PHONEME_SIL,                         // 40
+    TALKBACK_PHONEME_LAST = TALKBACK_PHONEME_SIL, // 40
+    TALKBACK_NUM_PHONEMES                         // 41 (0..40)
+};
+
+// -----------------------------------------------------------------------------
+// Function declarations.
+
+// ---------------------------
+// Startup/shutdown functions.
+// ---------------------------
+
+// Must be the first function called when using TalkBack.
+TALKBACK_ERR                   // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackStartupLibrary(
+    char const *iCoreDataDir); // IN:      full path of folder containing TalkBack data files.
+
+// Should be the last function called when using TalkBack.
+TALKBACK_ERR               // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackShutdownLibrary(); // IN:      nothing.
+
+// ------------------
+// Version functions.
+// ------------------
+
+// Gets the TalkBack version number.
+TALKBACK_ERR          // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetVersion(
+    long *oMajor,     // OUT:     major version number.
+    long *oMinor,     // OUT:     minor version number.
+    long *oRevision); // OUT:     revision version number.
+
+// Gets the TalkBack version number as a string.
+TALKBACK_ERR         // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetVersionString(
+    long  iMaxChars, // IN:      size of version string buffer.
+    char *oVersion); // OUT:     version string buffer.
+
+// ------------------
+// Utility functions.
+// ------------------
+
+// Checks whether a sound file can be analyzed and returns some quality metrics.
+//
+// NOTE: this function is deprecated and has been supplanted by
+// TalkBackGetSoundFileMetrics().
+TALKBACK_ERR                    // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackCheckSoundFile(
+    char const *iSoundFileName, // IN:      name of sound file to be checked.
+    long       *oCanBeAnalyzed, // OUT:     1 if sound can be analyzed, 0 if not.
+    long       *oIsClipped,     // OUT:     1 if sound is clipped, 0 if not.
+    double     *oDecibelRange); // OUT:     used decibel range of sound.
+
+// Returns metrics for the specified sound file.
+TALKBACK_ERR                                     // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetSoundFileMetrics(
+    char const                  *iSoundFileName, // IN:      name of sound file to be checked.
+    TALKBACK_SOUND_FILE_METRICS *ioMetrics);     // IN/OUT:  address of a structure where the metrics will be stored.
+
+// Checks whether text can be used for text-based analysis, returning the text
+// as it will be analyzed.
+TALKBACK_ERR                    // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackCheckSpokenText(
+    char const *iSpokenText,    // IN:      text to check.
+    long        iMaxChars,      // IN:      size of analyzed text buffer.
+    char       *oAnalyzedText); // OUT:     buffer for text as it will be analyzed.
+
+// Convert a TalkBack error code to a description string.
+TALKBACK_ERR                    // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetErrorString(
+    TALKBACK_ERR iErrorCode,    // IN:      TalkBack error code to convert.
+    long         iMaxChars,     // IN:      size of the buffer.
+    char        *oErrorString); // OUT:     buffer for the description string.
+
+// Gets the error code and text for the most recent TalkBack error.
+TALKBACK_ERR                     // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetLastError(
+    long          iMaxChars,    // IN:      size of the buffer.
+    char         *oErrorString, // OUT:     buffer for the description string.
+    TALKBACK_ERR *oErrorCode);  // OUT:     most recent TalkBack error code.
+
+// -------------------
+// Analysis functions.
+// -------------------
+
+// Gets an opaque TALKBACK_ANALYSIS object. This object is then queried with the
+// TalkBackGet* functions below.
+TALKBACK_ERR                                    // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetAnalysis(
+    TALKBACK_ANALYSIS         **ioAnalysis,     // IN/OUT:  address of a TALKBACK_ANALYSIS *variable where analysis will be stored.
+    char const                 *iSoundFileName, // IN:      name of the sound file to analyze.
+    char const                 *iSoundText,     // IN:      text spoken in sound file (can be NULL to use textless analysis).
+    TALKBACK_ANALYSIS_SETTINGS *iSettings);     // IN:      pointer to a TALKBACK_ANALYSIS_SETTINGS structure (can be NULL for defaults).
+
+// Frees an opaque TALKBACK_ANALYSIS object. This releases all memory used by
+// the analysis.
+TALKBACK_ERR                         // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackFreeAnalysis(
+    TALKBACK_ANALYSIS **ioAnalysis); // IN/OUT:  analysis to free.
+
+// #######################################################################
+// NOTE: all functions from this point on require a valid analysis object.
+// #######################################################################
+
+// ------------------------
+// Speech target functions.
+// ------------------------
+
+// Gets the number of speech target tracks.
+TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetNumSpeechTargetTracks(
+    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long              *oResult);  // OUT:     number of speech target tracks.
+
+// Gets the number of keys in the specified speech target track.
+TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetNumSpeechTargetKeys(
+    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long               iTrackNum, // IN:      speech target track.
+    long              *oResult);  // OUT:     number of keys in the speech target track.
+
+// Gets key information (time, value, derivative in, and derivative out) for the
+// specified key in the specified speech target track.
+TALKBACK_ERR                            // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetSpeechTargetKeyInfo(
+    TALKBACK_ANALYSIS *iAnalysis,       // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long               iTrackNum,       // IN:      speech target track.
+    long               iKeyNum,         // IN:      speech target key.
+    double            *oTime,           // OUT:     time of key.
+    double            *oValue,          // OUT:     value of key.
+    double            *oDerivativeIn,   // OUT:     incoming derivative of key.
+    double            *oDerivativeOut); // OUT:     outgoing derivative of key.
+
+// Gets the value of the function curve for the specified speech target track at
+// the specified time.
+TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetSpeechTargetValueAtTime(
+    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long               iTrackNum, // IN:      speech target track.
+    double             iTime,     // IN:      time in seconds.
+    double            *oResult);  // OUT:     value of the function curve.
+
+// Gets the derivatives of the function curve for the specified speech target
+// track at the specified time.
+TALKBACK_ERR                            // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetSpeechTargetDerivativesAtTime(
+    TALKBACK_ANALYSIS *iAnalysis,       // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long               iTrackNum,       // IN:      speech target track.
+    double             iTime,           // IN:      time in seconds.
+    double            *oDerivativeIn,   // OUT:     value of the incoming derivative of the function curve.
+    double            *oDerivativeOut); // OUT:     value of the outgoing derivative of the function curve.
+
+// -------------------------
+// Speech gesture functions.
+// -------------------------
+
+// Gets the number of speech gesture tracks.
+TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetNumGestureTracks(
+    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long              *oResult);  // OUT:     number of speech gesture tracks
+
+// Gets the number of keys in the specified speech gesture track.
+TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetNumGestureKeys(
+    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long               iTrackNum, // IN:      speech gesture track.
+    long              *oResult);  // OUT:     number of keys in the speech gesture track.
+
+// Gets key information (time, value, derivative in, and derivative out) for the
+// specified key in the specified speech gesture track.
+TALKBACK_ERR                            // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetGestureKeyInfo(
+    TALKBACK_ANALYSIS *iAnalysis,       // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long               iTrackNum,       // IN:      speech gesture track.
+    long               iKeyNum,         // IN:      speech gesture key.
+    double            *oTime,           // OUT:     time of key.
+    double            *oValue,          // OUT:     value of key.
+    double            *oDerivativeIn,   // OUT:     incoming derivative of key.
+    double            *oDerivativeOut); // OUT:     outgoing derivative of key.
+
+// Gets the value of the function curve for the specified speech gesture track
+// at the specified time.
+TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetGestureValueAtTime(
+    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long               iTrackNum, // IN:      speech gesture track.
+    double             iTime,     // IN:      time in seconds.
+    double            *oResult);  // OUT:     value of the function curve.
+
+// Gets the derivatives of the function curve for the specified speech gesture
+// track at the specified time.
+TALKBACK_ERR                            // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetGestureDerivativesAtTime(
+    TALKBACK_ANALYSIS *iAnalysis,       // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long               iTrackNum,       // IN:      speech gesture track.
+    double             iTime,           // IN:      time in seconds.
+    double            *oDerivativeIn,   // OUT:     value of the incoming derivative of the function curve.
+    double            *oDerivativeOut); // OUT:     value of the outgoing derivative of the function curve.
+
+// ----------------
+// Frame functions.
+// ----------------
+
+// NOTE: these functions use the frame rate specified in the
+// TALKBACK_ANALYSIS_SETTINGS structure passed to TalkBackGetAnalysis() and
+// default to 30 fps (TALKBACK_DEFAULT_FRAME_RATE) if the structure pointer was
+// NULL.
+
+// Gets the first frame number.
+TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetFirstFrameNum(
+    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long              *oResult);  // OUT:     number of the first frame.
+
+// Gets the last frame number.
+TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetLastFrameNum(
+    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long              *oResult);  // OUT:     number of the last frame.
+
+// Gets the start time of the specified frame.
+TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetFrameStartTime(
+    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long               iFrameNum, // IN:      frame.
+    double            *oResult);  // OUT:     start time of the frame in seconds.
+
+// Gets the end time of the specified frame.
+TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetFrameEndTime(
+    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long               iFrameNum, // IN:      frame.
+    double            *oResult);  // OUT:     end time of the frame in seconds.
+
+// Gets the value of the function curve for a speech target integrated over the
+// specified frame.
+TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetSpeechTargetValueAtFrame(
+    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long               iTrackNum, // IN:      speech target track.
+    long               iFrameNum, // IN:      frame number.
+    double            *oResult);  // OUT:     value of the function curve integrated over the frame.
+
+// Gets the dominant speech target at the specified frame.
+//
+// NOTE: this function is meant to be used in flipbook mode only.
+TALKBACK_ERR                                // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetDominantSpeechTargetAtFrame(
+    TALKBACK_ANALYSIS      *iAnalysis,      // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long                    iFrameNum,      // IN:      frame number.
+    TALKBACK_SPEECH_TARGET *oSpeechTarget); // OUT:     dominant speech target.
+
+// Gets the value of the function curve for a speech gesture integrated over the
+// specified frame.
+TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetGestureValueAtFrame(
+    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long               iTrackNum, // IN:      speech gesture track.
+    long               iFrameNum, // IN:      frame number.
+    double            *oResult);  // OUT:     value of the function curve integrated over the frame.
+
+// ------------------
+// Phoneme functions.
+// ------------------
+
+// Gets the number of phonemes.
+TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetNumPhonemes(
+    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long              *oResult);  // OUT:     number of phonemes.
+
+// Gets the enumeration of the specified phoneme.
+TALKBACK_ERR                        // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetPhonemeEnum(
+    TALKBACK_ANALYSIS *iAnalysis,   // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long               iPhonemeNum, // IN:      phoneme.
+    TALKBACK_PHONEME  *oResult);    // OUT:     enumeration of the specified phoneme.
+
+// Gets the start time of the specified phoneme.
+TALKBACK_ERR                        // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetPhonemeStartTime(
+    TALKBACK_ANALYSIS *iAnalysis,   // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long               iPhonemeNum, // IN:      phoneme.
+    double            *oResult);    // OUT:     start time of the phoneme in seconds.
+
+// Gets the end time of the specified phoneme.
+TALKBACK_ERR                        // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetPhonemeEndTime(
+    TALKBACK_ANALYSIS *iAnalysis,   // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long               iPhonemeNum, // IN:      phoneme.
+    double            *oResult);    // OUT:     end time of the phoneme in seconds.
+
+// ---------------
+// Word functions.
+// ---------------
+
+// NOTE: these functions only yield data for text-based analysis.
+
+// Gets the number of words.
+TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetNumWords(
+    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long              *oResult);  // OUT:     number of words.
+
+// Gets the text of the specified word.
+TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetWord(
+    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long               iWordNum,  // IN:      word.
+    long               iMaxChars, // IN:      size of word buffer.
+    char              *oWord);    // OUT:     word buffer.
+
+// Gets the start time of the specified word.
+TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetWordStartTime(
+    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long               iWordNum,  // IN:      word.
+    double            *oResult);  // OUT:     start time of the word in seconds.
+
+// Gets the end time of the specified word.
+TALKBACK_ERR                      // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackGetWordEndTime(
+    TALKBACK_ANALYSIS *iAnalysis, // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long               iWordNum,  // IN:      word.
+    double            *oResult);  // OUT:     end time of the word in seconds.
+
+// --------------------------
+// Phoneme editing functions.
+// --------------------------
+
+// Use these functions to modify the phoneme list after you get an opaque
+// analysis object from TalkBackGetAnalysis(). After modifying the phoneme list
+// in the opaque analysis object, subsequent TalkBackGet* calls on that opaque
+// analysis object for speech target (lip-synching) data will return values
+// based on the modified phoneme list. However, speech gesture data is not
+// affected by phoneme editing.
+//
+// NOTE: phoneme editing is only provided in order to support Ventriloquist-like
+// applications where tweaking of the phoneme segmenation (and subsequent
+// recalculation of the animation data) is required. Most customers probably
+// won't need this functionality.
+
+// Inserts a phoneme at the specified position in the specified manner.
+TALKBACK_ERR                            // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackInsertPhoneme(
+    TALKBACK_ANALYSIS *iAnalysis,       // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    TALKBACK_PHONEME   iPhoneme,        // IN:      enumeration of phoneme to insert.
+    long               iInsertPosition, // IN:      position (phoneme number) at which to insert.
+    int                iInsertBefore);  // IN:      manner of insertion:
+                                        //            0 means put phoneme after insert position;
+                                        //            1 means put phoneme before insert position.
+
+// Deletes the specified phoneme.
+TALKBACK_ERR                              // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackDeletePhoneme(
+    TALKBACK_ANALYSIS *iAnalysis,         // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long               iPhonemeToDelete); // IN:      phoneme to delete.
+
+// Changes the start time of the specified phoneme.
+//
+// NOTE: the start time specified may not be the actual start time for a number
+// of reasons, most notably if the specified start time will make the phoneme
+// too short. This function returns the actual start time so the caller can
+// check the result without having to query the phoneme.
+TALKBACK_ERR                             // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackChangePhonemeStart(
+    TALKBACK_ANALYSIS *iAnalysis,        // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long               iPhonemeToChange, // IN:      phoneme to change.
+    double            *ioNewTime);       // IN/OUT:  new start time value in seconds (in); actual start time (out).
+
+// Changes the end time of the specified phoneme.
+//
+// NOTE: the end time specified may not be the actual end time for a number of
+// reasons, most notably if the specified end time will make the phoneme too
+// short. This function returns the actual end time so the caller can check the
+// result without having to query the phoneme.
+TALKBACK_ERR                             // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackChangePhonemeEnd(
+    TALKBACK_ANALYSIS *iAnalysis,        // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long               iPhonemeToChange, // IN:      phoneme to change.
+    double            *ioNewTime);       // IN/OUT:  new end time value in seconds (in); actual end time (out).
+
+// Changes the enumeration of the specified phoneme.
+TALKBACK_ERR                             // RETURNS: TALKBACK_NOERR if successful, TalkBack error code if not.
+TalkBackChangePhonemeEnum(
+    TALKBACK_ANALYSIS *iAnalysis,        // IN:      opaque analysis object returned by TalkBackGetAnalysis().
+    long               iPhonemeToChange, // IN:      phoneme to change.
+    TALKBACK_PHONEME   iNewPhoneme);     // IN:      new phoneme enumeration.
+
+#if defined(__cplusplus)
+}
+#endif
+
+#endif