int inithtk(int argc, char *argv[],const char * app_version, bool noGraphics)
{
InitThreads(HT_MSGMON); // enable msg driven monitoring
if(InitShell(argc,argv,app_version)<SUCCESS) return FAIL;
// CommonInit() expanded with incompatible things commented out:
InitMem(); InitLabel();
InitMath(); InitSigP(); InitUtil();
InitWave(); /*InitAudio();*/ InitModel();
if(InitParm()<SUCCESS) return FAIL;
InitGraf((Boolean)noGraphics);
InitDict(); InitNet(); InitLM();
InitRec(); InitLat(); /*InitAdapt();*/
EnableBTrees(); /* allows unseen triphones to be synthesised */
/*
Register(abuffer_version);
Register(acode_version);
Register(acomponent_version);
Register(adict_version);
Register(agram_version);
Register(angram_version);
Register(ahmms_version);
Register(ahtk_version);
Register(amonitor_version);
Register(apacket_version);
Register(arec_version);
Register(aresource_version);
Register(arman_version);
Register(asource_version);
*/
InfoPrinted();
return SUCCESS;
}
#include "LWMap.h"
#include "LGBase.h"
/* -------------------------- Trace Flags ------------------------ */
static int trace = 0;
#define T_TOP 0001 /* Top Level tracing */
/* ---------------------- Global Variables ----------------------- */
static WordMap wmap; /* and the word map */
/* ---------------- Configuration Parameters --------------------- */
static ConfParam *cParm[MAXGLOBS];
static int nParm = 0; /* total num params */
/* ---------------- Process Command Line ------------------------- */
/* SetConfParms: set conf parms relevant to this tool */
void SetConfParms(void)
{
int i;
nParm = GetConfig("LGLIST", TRUE, cParm, MAXGLOBS);
if (nParm>0){
if (GetConfInt(cParm,nParm,"TRACE",&i)) trace = i;
int main(int argc, char *argv[])
{
char *s;
void Initialise(void);
InitShell(argc,argv,lfof_version,lfof_vc_id);
InitMem();
InitMath();
InitWave();
InitLabel();
InitWMap();
InitGBase();
InitPCalc();
if (!InfoPrinted() && NumArgs() == 0)
ReportUsage();
if (NumArgs() == 0) Exit(EXIT_SUCCESS);
SetConfParms();
while (NextArg() == SWITCHARG) {
s = GetSwtArg();
if (strlen(s)!=1)
HError(16719,"Bad switch %s; must be single letter",s);
switch(s[0]){
case 'f':
fofSize = GetChkedInt(1, 1000, s); break;
case 'n':
nSize = GetChkedInt(1, MAXNG, s); break;
case 'T':
trace = GetChkedInt(0,077,s); break;
default:
HError(16719,"LFoF: Unknown switch %s",s);
}
}
if (NextArg() != STRINGARG)
HError(16719,"LFoF: map file name expected");
mapFN = GetStrArg();
if (NextArg() != STRINGARG)
HError(16719,"LFoF: FoF file name expected");
fofFN = GetStrArg();
Initialise();
if (trace&T_TOP) {
printf("Calculating FoF table\n"); fflush(stdout);
}
fofTab = CreateFoFTab(&gstack,fofSize,nSize);
ComputeFoFTab(fofTab,nSize,&inSet);
WriteFoFTab(fofFN,fofTab,NULL);
Exit(EXIT_SUCCESS);
return EXIT_SUCCESS; /* never reached -- make compiler happy */
}
void Initialise(int argc, char *argv[])
{
InitThreads(HT_MSGMON);
if(InitShell(argc,argv,hnettest_version)<SUCCESS)
HError(999,"Initialise: couldnt init HShell");
InitMem(); InitLabel();
InitMath(); InitSigP();
InitWave(); InitAudio();
if(InitParm()<SUCCESS)
HError(999,"Initialise: couldnt init HParm");
InitGraf(FALSE); InitModel();
InitDict(); InitNet();
EnableBTrees(); /* allows unseen triphones to be synthesised */
if (!InfoPrinted() && NumArgs() == 0)
ReportUsage();
}
int main(int argc, char *argv[])
{
int i;
char *c,*s,*fn;
char sBuf[256],fmt[256];
void Initialise(void);
void ProcessText(char *fn,bool lastFile);
bool Exists(char *fn);
BackOffLM *CombineModels(MemHeap *heap,LMInfo *lmi,int nLModel,int nSize,WordMap *wl) ;
InitShell(argc,argv,ladapt_version,ladapt_vc_id);
InitMem();
InitMath();
InitWave();
InitLabel();
InitLUtil();
InitWMap();
InitGBase();
InitLModel();
InitPCalc();
InitPMerge();
SetConfParms();
if (!InfoPrinted() && NumArgs() == 0)
ReportUsage();
if (NumArgs() == 0) Exit(EXIT_SUCCESS);
InitBuildInfo(&binfo);
binfo.dctype = DC_ABSOLUTE;
nLModel = 1;
while (NextArg() == SWITCHARG) {
s = GetSwtArg();
if (strlen(s)!=1)
HError(16419,"Bad switch %s; must be single letter",s);
switch(s[0]){
case 'a':
newWords = GetChkedInt(10,10000000,s); break;
case 'b':
ngbSize = GetChkedInt(10,10000000,s); break;
case 'c':
i = GetChkedInt(2,LM_NSIZE,s);
binfo.cutOff[i] = GetChkedInt(0,1000,s);
break;
case 'd':
if (NextArg()!=STRINGARG)
HError(16419,"Gram base root file name expected");
rootFN = GetStrArg();
break;
case 'f':
strcpy(fmt, GetStrArg());
for (c=fmt; *c; *c=toupper(*c), c++); /* To uppercase */
if (strcmp(fmt, LM_TXT_TEXT)==0)
binfo.saveFmt = LMF_TEXT;
else if (strcmp(fmt, LM_TXT_BINARY)==0)
binfo.saveFmt = LMF_BINARY;
else if (strcmp(fmt, LM_TXT_ULTRA)==0)
binfo.saveFmt = LMF_ULTRA;
else
HError(16419,"Unrecognised LM format, should be one of [%s, %s, %s]",
LM_TXT_TEXT, LM_TXT_BINARY, LM_TXT_ULTRA);
break;
case 'g':
processText = FALSE; break;
case 'i':
if (NextArg()!=FLOATARG)
HError(16419,"Interpolation weight expected");
lmInfo[nLModel].weight = GetChkedFlt(0.0,1.0,s);
if (NextArg()!=STRINGARG)
HError(16419,"Interpolation LM filename expected");
lmInfo[nLModel].fn = GetStrArg();
nLModel++;
break;
case 'j':
i = GetChkedInt(2,LM_NSIZE,s);
binfo.wdThresh[i] = GetChkedFlt(0.0,1E10,s);
break;
case 'n':
nSize = GetChkedInt(1, MAXNG, s); break;
#ifdef HTK_TRANSCRIBER
case 's':
if (NextArg()!=STRINGARG)
HError(16419,"Gram file text source descriptor expected");
txtSrc = GetStrArg(); break;
case 't':
binfo.dctype = DC_KATZ; break;
#endif
case 'w':
if (NextArg()!=STRINGARG)
HError(16419,"Word list file name expected");
wlistFN = GetStrArg(); break;
#ifndef HTK_TRANSCRIBER
case 'x':
binfo.ptype = LMP_COUNT; break;
#endif
case 'T':
trace = GetChkedInt(0,077,s); break;
default:
HError(16419,"LAdapt: Unknown switch %s",s);
}
}
#ifdef HTK_TRANSCRIBER
if (nLModel==1) { /* must interpolate with at least one model */
HError(16419,"LAdapt: at least one model must be specified with -i option");
}
if (binfo.saveFmt==LMF_TEXT) { /* save fomat cannot be TEXT */
binfo.saveFmt=LMF_BINARY;
}
#endif
if (NextArg() != STRINGARG)
HError(16419,"LAdapt: language model file name expected");
outFN = CopyString(&gstack,GetStrArg());
Initialise();
if (processText) {
if (NextArg() != STRINGARG)
ProcessText(NULL,TRUE); /* input from stdin */
else
while (NextArg() == STRINGARG) {
/* !! copy string argument since it gets overwritten
by NextArg() when reading from script file */
fn = CopyString(&gstack,GetStrArg());
ProcessText(fn,NextArg() != STRINGARG);
}
if (NumArgs() != 0)
HError(-16419,"LAdapt: unused args left on cmd line");
for (i=0; i<stdBuf.ngb->fndx; i++) {
sprintf(sBuf,"%s.%d",stdBuf.ngb->fn,i);
AddInputGFile(&inSet,sBuf,1.0);
}
ResetHeap(&langHeap);
} else {
for (i=0; i<MAX_NGRAM_FILES; i++) {
sprintf(sBuf,"%s.%d",rootFN,i);
if (!Exists(sBuf))
break;
AddInputGFile(&inSet,sBuf,1.0);
}
if (i==MAX_NGRAM_FILES)
{
HError(-16419, "LAdapt: Only %d n-gram files read (recompile with different setting\nof MAX_NGRAM_FILES");
}
}
if (nLModel==1) {
adpLM = GenerateModel(&langHeap,&binfo);
} else {
if (binfo.ptype==LMP_COUNT)
binfo.ptype = LMP_FLOAT;
newLM = GenerateModel(&langHeap,&binfo);
lmInfo[0].lm = newLM;
lmInfo[0].fn = "unknown";
/* combine all models into one */
adpLM = CombineModels(&langHeap,lmInfo,nLModel,nSize,tgtVoc);
}
#ifdef HTK_TRANSCRIBER
#ifdef HTK_CRYPT
adpLM->encrypt = TRUE; /* force to write encrypted model */
#endif
#endif
SaveLangModel(outFN,adpLM);
Exit(EXIT_SUCCESS);
return EXIT_SUCCESS; /* never reached -- make compiler happy */
}
int main(int argc, char *argv[])
{
char *datafn, *s;
int stream = 0;
void Initialise(char *datafn);
void LoadFile(char *fn);
void CalcMeanCov(Sequence seq[], int s);
void ClusterVecs(Sequence seq[], int s);
void WriteVQTable(ClusterSet *cs[], char *fn);
if(InitShell(argc,argv,hquant_version,hquant_vc_id)<SUCCESS)
HError(2500,"HQuant: InitShell failed");
InitMem(); InitLabel();
InitMath(); InitSigP();
InitWave(); InitAudio();
InitVQ(); InitModel();
if(InitParm()<SUCCESS)
HError(2500,"HQuant: InitParm failed");
InitTrain();
if (!InfoPrinted() && NumArgs() == 0)
ReportUsage();
if (NumArgs() == 0) Exit(0);
SetConfParms();
InitStreamVars();
while (NextArg() == SWITCHARG) {
s = GetSwtArg();
if (strlen(s)!=1)
HError(2519,"HQuant: Bad switch %s; must be single letter",s);
switch(s[0]){
case 'd':
if ( ck != NULLC) HError(2519,"HQuant: Specify one of -d or -f, not both");
ck = INVDIAGC;
break;
case 'f':
if ( ck != NULLC) HError(2519,"HQuant: Specify one of -d or -f, not both");
ck = FULLC;
break;
case 'g':
globClustVar = TRUE;
break;
case 'l':
if (NextArg() != STRINGARG)
HError(2519,"HQuant: Segment label expected");
segLab = GetStrArg();
break;
case 'n':
if (NextArg() != INTARG)
HError(2519,"HQuant: Stream number expected");
stream = GetChkedInt(1,SMAX,s);
if (NextArg() != INTARG)
HError(2519,"HQuant: Codebook size expected");
cbSizes[stream]= GetChkedInt(1,32768,s);
break;
case 's':
if (NextArg() != INTARG)
HError(2519,"HQuant: Number of streams expected");
swidth[0] = GetChkedInt(1,SMAX,s);
break;
case 't':
tType = binTree;
break;
case 'w':
if (NextArg() != INTARG)
HError(2519,"HQuant: Stream number expected");
stream = GetChkedInt(1,SMAX,s);
if(swidth[0] < stream) swidth[0] = stream;
widthSet = TRUE;
if (NextArg() != INTARG)
HError(2519,"HQuant: Stream width expected");
swidth[stream]= GetChkedInt(1,256,s);
break;
case 'F':
if (NextArg() != STRINGARG)
HError(2519,"HQuant: Data File format expected");
if((dff = Str2Format(GetStrArg())) == ALIEN)
HError(-2589,"HQuant: Warning ALIEN Data file format set");
break;
case 'G':
if (NextArg() != STRINGARG)
HError(2519,"HQuant: Label File format expected");
if((lff = Str2Format(GetStrArg())) == ALIEN)
HError(-2589,"HQuant: Warning ALIEN Label file format set");
break;
case 'I':
if (NextArg() != STRINGARG)
HError(2519,"HQuant: MLF file name expected");
LoadMasterFile(GetStrArg());
break;
case 'L':
if (NextArg()!=STRINGARG)
HError(2519,"HQuant: Label file directory expected");
labDir = GetStrArg();
break;
case 'T':
if (NextArg() != INTARG)
HError(2519,"HQuant: Trace value expected");
trace = GetChkedInt(0,077,s);
break;
case 'X':
if (NextArg()!=STRINGARG)
HError(2519,"HQuant: Label file extension expected");
labExt = GetStrArg();
break;
default:
HError(2519,"HQuant: Unknown switch %s",s);
}
}
if (NextArg()!=STRINGARG)
HError(2519,"HQuant: Output VQ table file name expected");
vqfn = GetStrArg();
if (NextArg()!=STRINGARG)
HError(2519,"HQuant: Training data file name expected");
datafn = GetStrArg();
Initialise(datafn);
LoadFile(datafn);
while (NumArgs()>0) {
if (NextArg()!=STRINGARG)
HError(2519,"HQuant: Training data file name expected");
datafn = GetStrArg();
LoadFile(datafn);
}
for (stream=1;stream<=swidth[0];stream++){
if (trace&T_TOP)
printf("%s-clustering data for stream %d (width %d)\n",
(tType==linTree)?"Flat":"Tree",stream,swidth[stream]);
CalcMeanCov(dSeq,stream);
ClusterVecs(dSeq,stream);
}
WriteVQTable(cs,vqfn);
ResetTrain();
ResetParm();
ResetModel();
ResetVQ();
ResetAudio();
ResetWave();
ResetSigP();
ResetMath();
ResetLabel();
ResetMem();
ResetShell();
Exit(0);
return (0); /* never reached -- make compiler happy */
}
/* SetConfParms: set conf parms relevant to HCompV */
void SetConfParms(void)
{
int i;
Boolean b;
double f;
char buf[MAXSTRLEN];
nParm = GetConfig("HEREST", TRUE, cParm, MAXGLOBS);
if (nParm>0) {
if (GetConfInt(cParm,nParm,"TRACE",&i)) trace = i;
if (GetConfFlt(cParm,nParm,"VARFLOORPERCENTILE",&f)) varFloorPercent = f;
if (GetConfBool(cParm,nParm,"SAVEBINARY",&b)) saveBinary = b;
if (GetConfBool(cParm,nParm,"BINARYACCFORMAT",&b)) ldBinary = b;
/* 2-model reestimation alignment model set */
if (GetConfStr(cParm,nParm,"ALIGNMODELMMF",buf)) {
strcpy(al_hmmMMF,buf); al_hmmUsed = TRUE;
}
if (GetConfStr(cParm,nParm,"ALIGNHMMLIST",buf)) {
strcpy(al_hmmLst,buf); al_hmmUsed = TRUE;
}
/* allow multiple individual model files */
if (GetConfStr(cParm,nParm,"ALIGNMODELDIR",buf)) {
strcpy(al_hmmDir,buf); al_hmmUsed = TRUE;
}
if (GetConfStr(cParm,nParm,"ALIGNMODELEXT",buf)) {
strcpy(al_hmmExt,buf); al_hmmUsed = TRUE;
}
if (GetConfStr(cParm,nParm,"ALIGNXFORMEXT",buf)) {
xfInfo.alXFormExt = CopyString(&hmmStack,buf);
}
if (GetConfStr(cParm,nParm,"ALIGNXFORMDIR",buf)) {
xfInfo.alXFormDir = CopyString(&hmmStack,buf);
}
if (GetConfStr(cParm,nParm,"INXFORMMASK",buf)) {
xfInfo.inSpkrPat = CopyString(&hmmStack,buf);
}
if (GetConfStr(cParm,nParm,"PAXFORMMASK",buf)) {
xfInfo.paSpkrPat = CopyString(&hmmStack,buf);
}
if (GetConfStr(cParm,nParm,"LABFILEMASK",buf)) {
labFileMask = (char*)malloc(strlen(buf)+1);
strcpy(labFileMask, buf);
}
if (GetConfStr(cParm,nParm,"UPDATEMODE",buf)) {
if (!strcmp (buf, "DUMP")) updateMode = UPMODE_DUMP;
else if (!strcmp (buf, "UPDATE")) updateMode = UPMODE_UPDATE;
else if (!strcmp (buf, "BOTH")) updateMode = UPMODE_BOTH;
else HError(2319, "Unknown UPDATEMODE specified (must be DUMP, UPDATE or BOTH)");
}
}
}
void ReportUsage(void)
{
printf("\nUSAGE: HERest [options] hmmList dataFiles...\n\n");
printf(" Option Default\n\n");
printf(" -a Use an input linear transform off\n");
printf(" -c f Mixture pruning threshold 10.0\n");
printf(" -d s dir to find hmm definitions current\n");
printf(" -h s set output speaker name pattern *.%%%%%%\n");
printf(" to s, optionally set input and parent patterns\n");
printf(" -l N set max files per speaker off\n");
printf(" -m N set min examples needed per model 3\n");
printf(" -o s extension for new hmm files as src\n");
printf(" -p N set parallel mode to N off\n");
printf(" -r Enable Single Pass Training... \n");
printf(" ...using two parameterisations off\n");
printf(" -s s print statistics to file s off\n");
printf(" -t f [i l] set pruning to f [inc limit] inf\n");
printf(" -u tmvwap update t)rans m)eans v)ars w)ghts tmvw\n");
printf(" a)daptation xform p)rior used \n");
printf(" s)semi-tied xform \n");
printf(" -v f set minimum variance to f 0.0\n");
printf(" -w f set mix weight floor to f*MINMIX 0.0\n");
printf(" -x s extension for hmm files none\n");
printf(" -z s Save all xforms to TMF file s TMF\n");
PrintStdOpts("BEFGHIJKLMSTX");
printf("\n\n");
}
void SetuFlags(void)
{
char *s;
s=GetStrArg();
uFlags=(UPDSet) 0;
while (*s != '\0')
switch (*s++) {
case 't': uFlags = (UPDSet) (uFlags+UPTRANS); break;
case 'm': uFlags = (UPDSet) (uFlags+UPMEANS); break;
case 'v': uFlags = (UPDSet) (uFlags+UPVARS); break;
case 'w': uFlags = (UPDSet) (uFlags+UPMIXES); break;
case 's': uFlags = (UPDSet) (uFlags+UPSEMIT); break;
case 'a': uFlags = (UPDSet) (uFlags+UPXFORM); break;
case 'p': uFlags = (UPDSet) (uFlags+UPMAP); break;
default: HError(2320,"SetuFlags: Unknown update flag %c",*s);
break;
}
}
/* ScriptWord: return next word from script */
char *ScriptWord(FILE *script, char *scriptBuf)
{
int ch,qch,i;
i=0; ch=' ';
while (isspace(ch)) ch = fgetc(script);
if (ch==EOF) {
scriptBuf=NULL;
return NULL;
}
if (ch=='\'' || ch=='"'){
qch = ch;
ch = fgetc(script);
while (ch != qch && ch != EOF) {
scriptBuf[i++] = ch;
ch = fgetc(script);
}
if (ch==EOF)
HError(5051,"ScriptWord: Closing quote missing in script file");
} else {
do {
scriptBuf[i++] = ch;
ch = fgetc(script);
}while (!isspace(ch) && ch != EOF);
}
scriptBuf[i] = '\0';
return scriptBuf;
}
void CheckUpdateSetUp()
{
AdaptXForm *xf;
int main(int argc, char *argv[])
{
Source src;
int tmpInt;
float tmpFlt;
char *accfn, *s;
void Initialise(char *hmmListFn);
void Interpolate(void);
void UpdateModels(void);
void MakeWtAccLists(void);
void AttachWtAccLists(void);
void StatReport(void);
if(InitShell(argc,argv,hsmooth_version,hsmooth_vc_id)<SUCCESS)
HError(2400,"HSmooth: InitShell failed");
InitMem(); InitLabel();
InitMath(); InitSigP();
InitWave(); InitAudio();
InitVQ(); InitModel();
if(InitParm()<SUCCESS)
HError(2400,"HSmooth: InitParm failed");
InitTrain(); InitUtil();
if (!InfoPrinted() && NumArgs() == 0)
ReportUsage();
if (NumArgs() == 0) Exit(0);
SetConfParms();
CreateHeap(&hmmStack,"HmmStore", MSTAK, 1, 1.0, 50000, 500000);
CreateHMMSet(&hset,&hmmStack,TRUE);
while (NextArg() == SWITCHARG) {
s = GetSwtArg();
if (strlen(s)!=1)
HError(2419,"HSmooth: Bad switch %s; must be single letter",s);
switch(s[0]){
case 'b':
epsilon = GetChkedFlt(0.0,1.0,s); break;
case 'c':
maxStep = GetChkedInt(1,1000,s); break;
case 'd':
if (NextArg()!=STRINGARG)
HError(2419,"HSmooth: HMM definition directory expected");
hmmDir = GetStrArg(); break;
case 'e':
if (NextArg()!=STRINGARG)
HError(2419,"HSmooth: HMM definition directory expected");
newDir = GetStrArg(); break;
case 'm':
minEgs = GetChkedInt(1,1000,s); break;
case 'o':
if (NextArg()!=STRINGARG)
HError(2419,"HSmooth: HMM file extension expected");
newExt = GetStrArg(); break;
case 's':
stats = TRUE;
if (NextArg()!=STRINGARG)
HError(2419,"HSmooth: Stats file name expected");
statFN = GetStrArg(); break;
case 'u':
SetuFlags(); break;
case 'v':
minVar = GetChkedFlt(0.0,10.0,s); break;
case 'w':
mixWeightFloor = MINMIX * GetChkedFlt(0.0,10000.0,s);
break;
case 'x':
if (NextArg()!=STRINGARG)
HError(2419,"HSmooth: HMM file extension expected");
hmmExt = GetStrArg(); break;
case 'B':
saveBinary=TRUE;
break;
case 'H':
if (NextArg() != STRINGARG)
HError(2419,"HSmooth: HMM macro file name expected");
AddMMF(&hset,GetStrArg());
break;
case 'M':
if (NextArg()!=STRINGARG)
HError(2419,"HSmooth: Output macro file directory expected");
newDir = GetStrArg();
break;
case 'T':
trace = GetChkedInt(0,0100000,s); break;
default:
HError(2419,"HSmooth: Unknown switch %s",s);
}
}
if (NextArg() != STRINGARG)
HError(2419,"HSmooth: file name of HMM list expected");
Initialise(GetStrArg());
do {
if (NextArg()!=STRINGARG)
HError(2419,"HSmooth: accumulator file name expected");
accfn = GetStrArg();
src=LoadAccs(&hset,accfn,uFlags);
ReadFloat(&src,&tmpFlt,1,ldBinary);
totalPr += (LogDouble)tmpFlt;
ReadInt(&src,&tmpInt,1,ldBinary);
totalT += tmpInt;
CloseSource(&src);
nBlk++;
MakeWtAccLists();
} while (NumArgs()>0);
AttachWtAccLists();
Interpolate();
if (stats) StatReport();
UpdateModels();
Exit(0);
return (0); /* never reached -- make compiler happy */
}
int main(int argc, char *argv[])
{
char *datafn, *s;
int nSeg;
void Initialise(void);
void LoadFile(char *fn);
void EstimateModel(void);
void SaveModel(char *outfn);
if(InitShell(argc,argv,hinit_version,hinit_vc_id)<SUCCESS)
HError(2100,"HInit: InitShell failed");
InitMem(); InitLabel();
InitMath(); InitSigP();
InitWave(); InitAudio();
InitVQ(); InitModel();
if(InitParm()<SUCCESS)
HError(2100,"HInit: InitParm failed");
InitTrain(); InitUtil();
if (!InfoPrinted() && NumArgs() == 0)
ReportUsage();
if (NumArgs() == 0) Exit(0);
SetConfParms();
CreateHMMSet(&hset,&gstack,FALSE);
while (NextArg() == SWITCHARG) {
s = GetSwtArg();
if (strlen(s)!=1)
HError(2119,"HInit: Bad switch %s; must be single letter",s);
switch(s[0]){
case 'e':
epsilon = GetChkedFlt(0.0,1.0,s); break;
case 'i':
maxIter = GetChkedInt(0,100,s); break;
case 'l':
if (NextArg() != STRINGARG)
HError(2119,"HInit: Segment label expected");
segLab = GetStrArg();
break;
case 'm':
minSeg = GetChkedInt(1,1000,s); break;
case 'n':
newModel = FALSE; break;
case 'o':
outfn = GetStrArg();
break;
case 'u':
SetuFlags(); break;
case 'v':
minVar = GetChkedFlt(0.0,10.0,s); break;
case 'w':
mixWeightFloor = MINMIX * GetChkedFlt(0.0,10000.0,s);
break;
case 'B':
saveBinary = TRUE;
break;
case 'F':
if (NextArg() != STRINGARG)
HError(2119,"HInit: Data File format expected");
if((dff = Str2Format(GetStrArg())) == ALIEN)
HError(-2189,"HInit: Warning ALIEN Data file format set");
break;
case 'G':
if (NextArg() != STRINGARG)
HError(2119,"HInit: Label File format expected");
if((lff = Str2Format(GetStrArg())) == ALIEN)
HError(-2189,"HInit: Warning ALIEN Label file format set");
break;
case 'H':
if (NextArg() != STRINGARG)
HError(2119,"HInit: HMM macro file name expected");
AddMMF(&hset,GetStrArg());
break;
case 'I':
if (NextArg() != STRINGARG)
HError(2119,"HInit: MLF file name expected");
LoadMasterFile(GetStrArg());
break;
case 'L':
if (NextArg()!=STRINGARG)
HError(2119,"HInit: Label file directory expected");
labDir = GetStrArg(); break;
case 'M':
if (NextArg()!=STRINGARG)
HError(2119,"HInit: Output macro file directory expected");
outDir = GetStrArg();
break;
case 'T':
if (NextArg() != INTARG)
HError(2119,"HInit: Trace value expected");
trace = GetChkedInt(0,01777,s);
break;
case 'X':
if (NextArg()!=STRINGARG)
HError(2119,"HInit: Label file extension expected");
labExt = GetStrArg(); break;
default:
HError(2119,"HInit: Unknown switch %s",s);
}
}
if (NextArg()!=STRINGARG)
HError(2119,"HInit: source HMM file name expected");
hmmfn = GetStrArg();
Initialise();
do {
if (NextArg()!=STRINGARG)
HError(2119,"HInit: training data file name expected");
datafn = GetStrArg();
LoadFile(datafn);
} while (NumArgs()>0);
nSeg = NumSegs(segStore);
if (nSeg < minSeg)
HError(2121,"HInit: Too Few Observation Sequences [%d]",nSeg);
if (trace&T_TOP) {
printf("%d Observation Sequences Loaded\n",nSeg);
fflush(stdout);
}
EstimateModel();
SaveModel(outfn);
if (trace&T_TOP)
printf("Output written to directory %s\n",
outDir==NULL?"current":outDir);
Exit(0);
return (0); /* never reached -- make compiler happy */
}
int main(int argc, char *argv[])
{
int i;
char *s,*c,*e;
InitShell(argc,argv,lplex_version,lplex_vc_id);
InitMem();
InitMath();
InitWave();
InitLabel();
InitWMap();
InitCMap();
InitLUtil();
InitLModel();
InitPCalc();
InitPMerge();
SetConfParms();
if (!InfoPrinted() && NumArgs() == 0)
ReportUsage();
if (NumArgs() == 0) Exit(EXIT_SUCCESS);
nLModel = 1;
for (i=1; i<=LM_NSIZE; i++) cutOff[i] = 0, wdThresh[i] = 0.0;
CreateHeap(&permHeap, "permHeap", MSTAK, 1, 1.0, 4000, 20000);
CreateHeap(&tempHeap, "tempHeap", MSTAK, 1, 1.0, 8000, 40000);
while (NextArg() == SWITCHARG) {
s = GetSwtArg();
if (strlen(s)!=1)
HError(16619,"Bad switch %s; must be single letter",s);
switch(s[0]){
case 'c':
i = GetChkedInt(2,LM_NSIZE,s);
cutOff[i] = GetChkedInt(1,1000,s);
break;
case 'd':
i = GetChkedInt(2,LM_NSIZE,s);
wdThresh[i] = GetChkedFlt(0.0,1E10,s);
break;
case 'e':
if (NextArg() != STRINGARG)
HError(16619,"LPlex: Eq Class Name Expected");
c = GetStrArg();
if (NextArg() != STRINGARG)
HError(16619,"LPlex: Eq Label Name Expected");
e = GetStrArg();
AddEquiv(c,e);
break;
case 'i':
if (NextArg()!=FLOATARG)
HError(16619,"LPlex: Interpolation weight expected");
lmInfo[nLModel].weight = GetChkedFlt(0.0,1.0,s);
if (NextArg()!=STRINGARG)
HError(16619,"LPlex: Interpolation LM filename expected");
lmInfo[nLModel].fn = GetStrArg();
nLModel++;
break;
case 'n':
testInfo[numTests++] = GetChkedInt(1, 10, s); break;
case 'o':
printOOV = TRUE; break;
case 's':
if (NextArg() != STRINGARG)
HError(16619,"LPlex: Prob Stream file name expected");
outStreamFN = GetStrArg();
break;
case 't':
streamMode = TRUE; break;
case 'u':
skipOOV = FALSE; break;
case 'w':
if (NextArg() != STRINGARG)
HError(16619,"LPlex: Word list file name expected");
wlistFN = GetStrArg();
break;
case 'z':
if (NextArg() != STRINGARG)
HError(16619,"LPlex: New null class name expected");
nulName = GetStrArg();
break;
case 'G':
if (NextArg() != STRINGARG)
HError(16619,"Label File format expected");
if((lff = Str2Format(GetStrArg())) == ALIEN)
HError(16619,"Warning ALIEN Label file format set");
break;
case 'I':
if (NextArg() != STRINGARG)
HError(16619,"MLF file name expected");
LoadMasterFile(GetStrArg()); break;
case 'T':
trace = GetChkedInt(0,077, s); break;
default:
HError(16619,"LPlex: Unknown switch %s",s);
}
}
#ifdef HTK_TRANSCRIBER
if (trace&T_PROB) trace=trace^T_PROB;
#endif
if (NextArg()!=STRINGARG) /* load the language model */
HError(16619, "Language model filename expected");
lmInfo[0].fn = GetStrArg();
Initialise();
ProcessFiles();
Exit(EXIT_SUCCESS);
return EXIT_SUCCESS; /* never reached -- make compiler happy */
}
int main(int argc, char *argv[])
{
int i;
char *s,*c;
char fmt[256];
dictList *dEntry,*d;
InitShell(argc,argv,prog_version,prog_vc_id);
InitMem();
InitMath();
InitWave();
InitLabel();
InitDict();
InitWMap();
InitLUtil();
InitLModel();
InitPCalc();
if (!InfoPrinted() && NumArgs() == 0)
ReportUsage();
if (NumArgs() == 0) Exit(EXIT_SUCCESS);
SetConfParms();
CreateHeap(&langHeap,"langHeap",MSTAK,1,0.5,5000,40000);
for (i=1; i<=LM_NSIZE; i++) cutOff[i] = 0;
while (NextArg() == SWITCHARG) {
s = GetSwtArg();
if (strlen(s)!=1)
HError(16919,"Bad switch %s; must be single letter",s);
switch(s[0]){
case 'c':
i = GetChkedInt(2,LM_NSIZE,s);
cutOff[i] = GetChkedInt(1,1000,s);
break;
case 'd':
if (NextArg()!=STRINGARG)
HError(16919,"LMCopy: Input dictionary file name expected");
dEntry=New(&gcheap,sizeof(dictList));
dEntry->fname=GetStrArg(); dEntry->next=NULL;
if (dList==NULL) dList=dEntry;
else {
for (d=dList;d->next!=NULL;d=d->next);
d->next=dEntry;
}
break;
case 'f':
strcpy(fmt,GetStrArg());
for (c=fmt; *c!=0; *c=toupper(*c), c++);
if (strcmp(fmt, LM_TXT_TEXT)==0)
saveFmt = LMF_TEXT;
else if (strcmp(fmt, LM_TXT_BINARY)==0)
saveFmt = LMF_BINARY;
else if (strcmp(fmt, LM_TXT_ULTRA)==0)
saveFmt = LMF_ULTRA;
else
HError(16919,"Unrecognised LM format, should be one of [%s, %s, %s]",
LM_TXT_TEXT, LM_TXT_BINARY, LM_TXT_ULTRA);
break;
case 'm':
remDup=FALSE;
break;
case 'n':
nSize = GetChkedInt(1,LM_NSIZE,s); break;
case 'o':
firstOnly=TRUE;
break;
case 'u':
if (NextArg()!=STRINGARG)
HError(16919,"LMCopy: Unigram file name expected");
uniFn = GetStrArg();
break;
case 'v':
if (NextArg()!=STRINGARG)
HError(16919,"LMCopy: Dictionary output file name expected");
outDictFn = GetStrArg();
break;
case 'w':
if (NextArg() != STRINGARG)
HError(16919,"LPlex: Word list file name expected");
wlistFN = GetStrArg();
break;
case 'T':
trace = GetChkedInt(0,077, s); break;
default:
HError(16919,"LMPlex: Unknown switch %s",s);
}
}
if (NextArg()!=STRINGARG) /* load the language model */
HError(16919, "Input language model filename expected");
srcFN = GetStrArg();
if (NextArg()!=STRINGARG) /* load the language model */
HError(16919, "Output language model filename expected");
tgtFN= GetStrArg();
if (wlistFN!=NULL) {
InitVocab(&vocab);
if(ReadDict(wlistFN,&vocab) < SUCCESS)
HError(16913,"Could not read dict in %s", wlistFN);
if (trace&T_TOP) {
printf("Loaded %d words from %s\n",vocab.nwords,wlistFN);
fflush(stdout);
}
voc = &vocab;
CreateWordList(wlistFN,&wlist,10);
lm = LoadLangModel(srcFN,&wlist,1.0,LMP_FLOAT|LMP_COUNT,&langHeap);
} else {
voc = NULL;
lm = LoadLangModel(srcFN,NULL,1.0,LMP_FLOAT|LMP_COUNT,&langHeap);
}
if (trace&T_TOP) {
printf("Loaded model from %s\n",srcFN);
fflush(stdout);
}
if (lm->probType==LMP_COUNT) {
RebuildLM(lm, cutOff, NULL, LMP_FLOAT); /* GLM there was no threshold before! */
}
if (uniFn!=NULL)
ReplaceUnigrams(uniFn,lm);
if (nSize>0 && nSize<lm->nSize)
lm->nSize = nSize;
#ifdef HTK_CRYPT
if (lm->encrypt && saveFmt==LMF_TEXT)
saveFmt = LMF_BINARY;
#endif
for (i=1;i<=lm->nSize;i++)
lm->gInfo[i].fmt = (i==1) ? LMF_TEXT : saveFmt;
SaveLangModel(tgtFN,lm);
if (trace&T_TOP) {
printf("Wrote model to %s\n",tgtFN);
fflush(stdout);
}
if (outDictFn) {
MakeDictionary(outDictFn,dList,voc);
}
Exit(EXIT_SUCCESS);
return EXIT_SUCCESS; /* never reached -- make compiler happy */
}
#include "HLM.h"
typedef enum {unknown, wordLoop, boBiGram, matBiGram, multiLat, wordPair} BuildType;
static int trace = 0; /* Trace flags */
static LabId enterId; /* id of !ENTRY label in ngram */
static LabId exitId; /* id of !EXIT label in ngram */
static LabId bStartId=NULL; /* id of start bracket */
static LabId bEndId=NULL; /* id of end bracket */
static LabId unknownId; /* id of unknown label in ngram */
static Boolean zapUnknown = FALSE; /* zap unknown symbols from bigram */
MemHeap buildStack;
/* ---------------- Configuration Parameters --------------------- */
static ConfParam *cParm[MAXGLOBS];
static int nParm = 0; /* total num params */
/* ---------------- Process Command Line ------------------------- */
/* SetConfParms: set conf parms relevant to this tool */
void SetConfParms(void)
{
int i;
nParm = GetConfig("HBUILD", TRUE, cParm, MAXGLOBS);
if (nParm>0){
if (GetConfInt(cParm,nParm,"TRACE",&i)) trace = i;
}
}
void ReportUsage(void)
{
printf("\nUSAGE: HBuild [options] wordList latFile\n\n");
printf(" Option Default\n\n");
printf(" -b binary lattice output ASCII\n");
printf(" -m s load matrix bigram from s off\n");
printf(" -n s load back-off bigram from s off\n");
printf(" -s s1 s2 s1/s2 are bigram start/end labels !ENTER !EXIT\n");
printf(" -t s1 s2 bracket word-loop/pair with s1 s2 off\n");
printf(" -u s set unknown symbol to s !NULL\n");
printf(" -w s load word-pair grammar from s off\n");
printf(" -x s load multi-level lattice from s off\n");
printf(" -z ignore ngrams with unknown symbol off\n");
PrintStdOpts("");
printf("\n\n");
}
int main(int argc, char *argv[])
{
char *wordListFn,*latFn,*ipFn=NULL;
LModel *bigramLm;
BuildType bType = unknown;
Boolean saveLatBin = FALSE;
LatFormat format = HLAT_LMLIKE;
Lattice *lat,*ipLat;
Vocab voc;
char *s;
Lattice *ProcessWordLoop(MemHeap *latHeap, Vocab *voc);
Lattice *ProcessBiGram(MemHeap *latHeap, Vocab *voc, LModel *biLM);
void SaveLattice(Lattice *lat, char *latFn, LatFormat format);
Lattice *LoadLattice(MemHeap *latHeap, char *latFn, Vocab *voc,
Boolean shortArc);
Lattice *ProcessWordPair(MemHeap *latHeap, Vocab *voc, char *fn);
if(InitShell(argc,argv,hbuild_version,hbuild_vc_id)<SUCCESS)
HError(3000,"HBuild: InitShell failed");
InitMem(); InitLabel();
InitMath();
InitDict(); InitNet();
InitLM();
CreateHeap(&buildStack, "HBuild Stack", MSTAK, 1, 0.0, 100000, LONG_MAX );
if (!InfoPrinted() && NumArgs() == 0)
ReportUsage();
if (NumArgs() == 0) Exit(0);
SetConfParms();
int main(int argc, char *argv[])
{
char *s,*lfn,*dfn;
void GenSentences(char *latfn, char *dicfn);
if(InitShell(argc,argv,hsgen_version,hsgen_vc_id)<SUCCESS)
HError(3400,"HSGen: InitShell failed");
InitMem(); InitLabel();
InitMath(); InitSigP();
InitWave(); InitAudio();
InitVQ(); InitModel();
if(InitParm()<SUCCESS)
HError(3200,"HSGen: InitParm failed");
InitDict();
InitNet();
if (!InfoPrinted() && NumArgs() == 0)
ReportUsage();
if (NumArgs() == 0) Exit(0);
SetConfParms();
while (NextArg() == SWITCHARG) {
s = GetSwtArg();
if (strlen(s)!=1)
HError(3419, "Bad switch %s; must be single letter", s);
switch(s[0]){
case 's':
stats = TRUE; break;
case 'l':
lnum = TRUE; break;
case 'n':
ngen = GetChkedInt(1,1000000,s); break;
case 'q':
quiet = TRUE; break;
case 'T':
trace = GetChkedInt(0,07,s); break;
default:
HError(3419, "Unknown switch %s", s);
}
}
if (NextArg()!=STRINGARG)
HError(3419, "lattice file name expected");
lfn = GetStrArg();
if (NextArg()!=STRINGARG)
HError(3419, "dictionary file name expected");
dfn = GetStrArg();
GenSentences(lfn,dfn);
if(InitShell(argc,argv,hsgen_version,hsgen_vc_id)<SUCCESS)
HError(3400,"HSGen: InitShell failed");
ResetNet();
ResetDict();
ResetParm();
ResetModel();
ResetVQ();
ResetAudio();
ResetWave();
ResetSigP();
ResetMath();
ResetLabel();
ResetMem();
ResetShell();
Exit(0);
return (0); /* never reached -- make compiler happy */
}
int main(int argc, char *argv[])
{
int i;
char *s,*c;
char fmt[256];
InitShell(argc,argv,lnorm_version,lnorm_vc_id);
InitMem();
InitMath();
InitWave();
InitLabel();
InitWMap();
InitLUtil();
InitLModel();
InitPCalc();
if (!InfoPrinted() && NumArgs() == 0)
ReportUsage();
if (NumArgs() == 0) Exit(EXIT_SUCCESS);
SetConfParms();
CreateHeap(&langHeap,"langHeap",MSTAK,1,0.5,5000,40000);
for (i=1; i<=LM_NSIZE; i++) cutOff[i] = 0, wdThresh[i] = 0.0;
while (NextArg() == SWITCHARG) {
s = GetSwtArg();
if (strlen(s)!=1)
HError(16519,"Bad switch %s; must be single letter",s);
switch(s[0]){
case 'c':
i = GetChkedInt(2,LM_NSIZE,s);
cutOff[i] = GetChkedInt(1,1000,s);
break;
case 'd':
i = GetChkedInt(2,LM_NSIZE,s);
wdThresh[i] = GetChkedFlt(0.0,1E10,s);
break;
case 'f':
strcpy(fmt,GetStrArg());
for (c=fmt; *c!=0; *c=toupper(*c),c++);
if (strcmp(fmt, LM_TXT_TEXT)==0)
saveFmt = LMF_TEXT;
else if (strcmp(fmt, LM_TXT_BINARY)==0)
saveFmt = LMF_BINARY;
else if (strcmp(fmt, LM_TXT_ULTRA)==0)
saveFmt = LMF_ULTRA;
else
HError(16519,"Unrecognised LM format, should be one of [%s, %s, %s]",
LM_TXT_TEXT, LM_TXT_BINARY, LM_TXT_ULTRA);
break;
case 'n':
nSize = GetChkedInt(1,LM_NSIZE,s); break;
case 'w':
if (NextArg() != STRINGARG)
HError(16519,"LPlex: Word list file name expected");
wlistFN = GetStrArg();
break;
case 'T':
trace = GetChkedInt(0,077, s); break;
default:
HError(16519,"LMPlex: Unknown switch %s",s);
}
}
if (NextArg()!=STRINGARG) /* load the language model */
HError(16519, "Input language model filename expected");
srcFN = GetStrArg();
if (NextArg()!=STRINGARG) /* load the language model */
HError(16519, "Output language model filename expected");
tgtFN= GetStrArg();
if (wlistFN!=NULL) {
CreateWordList(wlistFN,&wlist,10);
lm = LoadLangModel(srcFN,&wlist,1.0,LMP_FLOAT|LMP_COUNT,&langHeap);
} else {
lm = LoadLangModel(srcFN,NULL,1.0,LMP_FLOAT|LMP_COUNT,&langHeap);
}
if (lm->probType==LMP_COUNT) {
RebuildLM(lm,cutOff,wdThresh,LMP_FLOAT);
} else {
NormaliseLM(lm);
}
if (nSize>0 && nSize<lm->nSize)
lm->nSize = nSize;
for (i=1;i<=lm->nSize;i++)
lm->gInfo[i].fmt = (i==1) ? LMF_TEXT : saveFmt;
SaveLangModel(tgtFN,lm);
Exit(EXIT_SUCCESS);
return EXIT_SUCCESS; /* never reached -- make compiler happy */
}
strcpy(vName,"varFloor"); strcat(vName,num);
fprintf(f,"~v %s\n",vName);
if (fullcNeeded[s])
TriDiag2Vector(accs[s].squareSum.inv,v);
else
CopyVector(accs[s].fixed.var,v);
for (i=1; i<=hset.swidth[s]; i++)
v[i] *= vFloorScale;
fprintf(f,"<Variance> %d\n",hset.swidth[s]);
WriteVector(f,v,FALSE);
FreeVector(&gstack,v);
}
fclose(f);
if (trace&T_TOP)
printf("Var floor macros output to file %s\n",outfn);
}
/* ---------------- Load Data and Accumulate Stats --------------- */
/* AccVar: update global accumulators with given observation */
void AccVar(Observation obs)
{
int x,y,s,V;
float val;
Vector v;
totalCount++;
for (s=1; s<=hset.swidth[0]; s++){
v = obs.fv[s]; V = hset.swidth[s];
for (x=1;x<=V;x++) {
val=v[x];
accs[s].meanSum[x] += val; /* accumulate mean */
if (fullcNeeded[s]) { /* accumulate covar */
accs[s].squareSum.inv[x][x] += val*val;
for (y=1;y<x;y++)
accs[s].squareSum.inv[x][y] += val*v[y];
} else /* accumulate var */
accs[s].squareSum.var[x] += val*val;
}
}
}
/* CheckData: check data file consistent with HMM definition */
void CheckData(char *fn, BufferInfo info)
{
if (info.tgtVecSize!=hset.vecSize)
HError(2050,"CheckData: Vector size in %s[%d] is incompatible with hmm %s[%d]",
fn,info.tgtVecSize,hmmfn,hset.vecSize);
if (info.tgtPK != hset.pkind)
HError(2050,"CheckData: Parameterisation in %s is incompatible with hmm %s",
fn,hmmfn);
}
/* LoadFile: load whole file or segments and accumulate variance */
void LoadFile(char *fn)
{
ParmBuf pbuf;
BufferInfo info;
char labfn[80];
Transcription *trans;
long segStIdx,segEnIdx;
int i,j,ncas,nObs;
LLink p;
if (segId == NULL) { /* load whole parameter file */
if((pbuf=OpenBuffer(&iStack, fn, 0, dff, FALSE_dup, FALSE_dup))==NULL)
HError(2050,"LoadFile: Config parameters invalid");
GetBufferInfo(pbuf,&info);
CheckData(fn,info);
nObs = ObsInBuffer(pbuf);
for (i=0; i<nObs; i++){
ReadAsTable(pbuf,i,&obs);
AccVar(obs);
}
if (trace&T_LOAD) {
printf(" %d observations loaded from %s\n",nObs,fn);
fflush(stdout);
}
CloseBuffer(pbuf);
}
else { /* load segment of parameter file */
MakeFN(fn,labDir,labExt,labfn);
trans = LOpen(&iStack,labfn,lff);
ncas = NumCases(trans->head,segId);
if ( ncas > 0) {
if((pbuf=OpenBuffer(&iStack, fn, 0, dff, FALSE_dup, FALSE_dup))==NULL)
static Boolean tgtHdr = FALSE; /* print target header info */
static Boolean obsFmt = FALSE; /* print observation format */
static Boolean prData = TRUE; /* print data */
static Boolean rawOut = FALSE; /* raw output i.e no numbering */
static Boolean replay = FALSE; /* replay audio */
static Boolean frcDisc = FALSE; /* List VQ symbols from cont file */
static FileFormat ff = UNDEFF; /* Source File format */
static long gst = -1; /* start sample to list */
static long gen = -1; /* end sample to list */
static int numS = 1; /* number of streams */
static int nItems = 10; /* num items per line */
static int barwidth; /* width of printed bars */
static char barc = '-'; /* bar character */
/* ---------------- Configuration Parameters --------------------- */
static ConfParam *cParm[MAXGLOBS];
static int nParm = 0; /* total num params */
static HTime sampPeriod; /* raw audio input only */
static int audSignal;
/* ---------------- Process Command Line ------------------------- */
/* SetConfParms: set conf parms relevant to this tool */
void SetConfParms(void)
{
int i;
double d;
sampPeriod = 0.0; audSignal = NULLSIG;
nParm = GetConfig("HLIST", TRUE, cParm, MAXGLOBS);
if (nParm>0){
if (GetConfInt(cParm,nParm,"AUDIOSIG",&i)) audSignal = i;
if (GetConfInt(cParm,nParm,"TRACE",&i)) trace = i;
if (GetConfFlt(cParm,nParm,"SOURCERATE",&d)) sampPeriod = d;
}
}
int main_HCopy(int argc, char *argv[])
{
char *s; /* next file to process */
void OpenSpeechFile(char *s);
void AppendSpeechFile(char *s);
void PutTargetFile(char *s);
if(InitShell(argc,argv,hcopy_version,hcopy_vc_id)<SUCCESS)
HError(1000,"HCopy: InitShell failed");
InitMem(); InitLabel();
InitMath(); InitSigP();
InitWave(); InitAudio();
InitVQ(); InitModel();
if(InitParm()<SUCCESS)
HError(1000,"HCopy: InitParm failed");
if (!InfoPrinted() && NumArgs() == 0)
ReportUsageHCopy();
if (NumArgs() == 0) return(0);
SetConfParmsHCopy();
/* initial trace string is null */
trList.str = NULL;
CreateHeap(&iStack, "InBuf", MSTAK, 1, 0.0, STACKSIZE, LONG_MAX);
CreateHeap(&oStack, "OutBuf", MSTAK, 1, 0.0, STACKSIZE, LONG_MAX);
CreateHeap(&cStack, "ChopBuf", MSTAK, 1, 0.0, STACKSIZE, LONG_MAX);
CreateHeap(&lStack, "LabBuf", MSTAK, 1, 0.0, 10000, LONG_MAX);
CreateHeap(&tStack, "Trace", MSTAK, 1, 0.0, 100, 200);
while (NextArg() == SWITCHARG) {
s = GetSwtArg();
if (strlen(s)!=1)
HError(1019,"HCopy: Bad switch %s; must be single letter",s);
switch(s[0]){
case 'a':
if (NextArg() != INTARG)
HError(1019,"HCopy: Auxiliary label index expected");
auxLab = GetChkedInt(1,100000,s) - 1;
break;
case 'e': /* end time in seconds, max 10e5 secs */
en = GetChkedFlt(-MAXTIME,MAXTIME,s);
stenSet = TRUE; chopF = TRUE;
break;
case 'i':
if (NextArg() != STRINGARG)
HError(1019,"HCopy: Output MLF name expected");
if(SaveToMasterfile(GetStrArg())<SUCCESS)
HError(1014,"HCopy: Cannot write to MLF");
useMLF = TRUE; labF = TRUE; break;
case 'l':
if (NextArg() != STRINGARG)
HError(1019,"HCopy: Target label file directory expected");
outLabDir = GetStrArg();
labF = TRUE; break;
case 'm':
xMargin = GetChkedFlt(-MAXTIME,MAXTIME,s);
chopF = TRUE; break;
case 'n':
if (NextArg() != INTARG)
HError(1019,"HCopy: Label index expected");
labstidx= GetChkedInt(-100000,100000,s);
if (NextArg() == INTARG)
labenidx = GetChkedInt(-100000,100000,s);
chopF = TRUE; break;
case 's': /* start time in seconds */
st = GetChkedFlt(0,MAXTIME,s);
stenSet = TRUE; chopF = TRUE; break;
case 't':
if (NextArg() != INTARG)
HError(1019,"HCopy: Trace line width expected");
traceWidth= GetChkedInt(10,100000,s); break;
case 'x':
if (NextArg() != STRINGARG)
HError(1019,"HCopy: Label name expected");
labName = GetLabId(GetStrArg(),TRUE);
if (NextArg() == INTARG)
labRep = GetChkedInt(1,100000,s);
chopF = TRUE; labF = TRUE; break;
case 'F':
if (NextArg() != STRINGARG)
HError(1019,"HCopy: Source file format expected");
if((srcFF = Str2Format(GetStrArg())) == ALIEN)
HError(-1089,"HCopy: Warning ALIEN src file format set");
break;
case 'G':
if (NextArg() != STRINGARG)
HError(1019,"HCopy: Source label File format expected");
if((srcLabFF = Str2Format(GetStrArg())) == ALIEN)
HError(-1089,"HCopy: Warning ALIEN Label output file format set");
labF= TRUE; break;
case 'I':
if (NextArg() != STRINGARG)
HError(1019,"HCopy: MLF file name expected");
LoadMasterFile(GetStrArg());
labF = TRUE; break;
case 'L':
if (NextArg()!=STRINGARG)
HError(1019,"HCopy: Label file directory expected");
labDir = GetStrArg();
labF = TRUE; break;
case 'P':
if (NextArg() != STRINGARG)
HError(1019,"HCopy: Label File format expected");
if((tgtLabFF = Str2Format(GetStrArg())) == ALIEN)
HError(-1089,"HCopy: Warning ALIEN Label file format set");
labF = TRUE; break;
case 'O':
if (NextArg() != STRINGARG)
HError(1019,"HCopy: Target file format expected");
if((tgtFF = Str2Format(GetStrArg())) == ALIEN)
HError(-1089,"HCopy: Warning ALIEN target file format set");
break;
case 'T':
trace = GetChkedInt(0,16,s); break;
case 'X':
if (NextArg()!=STRINGARG)
HError(1019,"HCopy: Label file extension expected");
labExt = GetStrArg();
labF = TRUE; break;
default:
HError(1019,"HCopy: Unknown switch %s",s);
}
}
if (NumArgs() == 1)
HError(1019,"HCopy: Target file or + operator expected");
FixOptions();
while (NumArgs()>1) { /* process group S1 + S2 + ... TGT */
off = 0.0;
if (NextArg()!=STRINGARG)
HError(1019,"HCopy: Source file name expected");
s = GetStrArg();
OpenSpeechFile(s); /* Load initial file S1 */
if (NextArg()!=STRINGARG)
HError(1019,"HCopy: Target file or + operator expected");
s = GetStrArg();
while (strcmp(s,"+") == 0) { /* Append + S2 + S3 ... */
if (NextArg()!=STRINGARG)
HError(1019,"HCopy: Append file name expected");
s = GetStrArg();
AppendSpeechFile(s);
if (NextArg()!=STRINGARG)
HError(1019,"HCopy: Target file or + operator expected");
s = GetStrArg();
}
PutTargetFile(s);
if(trace & T_MEM) PrintAllHeapStats();
if(trans != NULL){
trans = NULL;
ResetHeap(&lStack);
}
ResetHeap(&iStack);
ResetHeap(&oStack);
if(chopF) ResetHeap(&cStack);
}
if(useMLF) CloseMLFSaveFile();
if (NumArgs() != 0) HError(-1019,"HCopy: Unused args ignored");
return (0); /* never reached -- make compiler happy */
}
static int nSize = 0; /* output ngram size */
static WordMap wList; /* the word list */
static int nLModel; /* number of loaded LMs */
static LMInfo lmInfo[MAX_LMODEL]; /* array of loaded LMs */
static BackOffLM *tgtLM; /* target lm */
static char *tgtFN; /* output model name */
static MemHeap langHeap; /* Stores global stats */
static LMFileFmt saveFmt = DEF_SAVEFMT; /* LM file format */
/* ---------------- Configuration Parameters --------------------- */
static ConfParam *cParm[MAXGLOBS];
static int nParm = 0; /* total num params */
/* ---------------- Function prototypes -------------------------- */
void Initialise(void);
/* ---------------- Process Command Line ------------------------- */
/* SetConfParms: set conf parms relevant to this tool */
void SetConfParms(void)
{
int i;
nParm = GetConfig("LMERGE", TRUE, cParm, MAXGLOBS);
if (nParm>0){
if (GetConfInt(cParm,nParm,"TRACE",&i)) trace = i;
}
}
char *ReturnLMName(int fmt)
{
switch(fmt) {
case LMF_TEXT:
return LM_TXT_TEXT;
case LMF_BINARY:
return LM_TXT_BINARY;
case LMF_ULTRA:
return LM_TXT_ULTRA;
default:
return LM_TXT_OTHER;
int main(int argc, char *argv[])
{
char * labFn, *listfn, *s;
int i,fidx;
MLFEntry *me = NULL;
Transcription *t;
void InitStats(char *listfn);
void GatherStats(Transcription *t);
void OutputStats(void);
if(InitShell(argc,argv,hlstats_version,hlstats_vc_id)<SUCCESS)
HError(1300,"HLStats: InitShell failed");
InitMem(); InitMath();
InitWave(); InitLabel();
InitLM();
if (!InfoPrinted() && NumArgs() == 0)
ReportUsage();
if (NumArgs() == 0) Exit(0);
SetConfParms();
enterId=GetLabId("!ENTER",TRUE); /* All sentences should or are coerced */
exitId=GetLabId("!EXIT",TRUE); /* to start enterId and end exitId */
nullId=GetLabId("!NULL",TRUE); /* Name for words not in list */
while (NextArg() == SWITCHARG) {
s = GetSwtArg();
if (strlen(s)!=1)
HError(1319,"HLStats: Bad switch %s; must be single letter",s);
switch(s[0]){
case 'b':
doBigram = TRUE;
if (NextArg() != STRINGARG)
HError(1319,"HLStats: Ngram output file name expected");
bigFile = GetStrArg();
break;
case 'c':
doLCount = TRUE;
lCountLimit = GetChkedInt(0,100000,s);
break;
case 'd':
doDurs = TRUE; break;
case 'f':
bigFloor = GetChkedFlt(0.0,1000.0,s);
break;
case 'h':
hSize = GetChkedInt(1,2,s);
break;
case 'l':
doList = TRUE;
if (NextArg() != STRINGARG)
HError(1319,"HLStats: Output label list file name expected");
listFile = GetStrArg();
break;
case 'o':
doBOff = TRUE;
break;
case 'p':
doPCount = TRUE;
pCountLimit = GetChkedInt(0,100000,s);
break;
case 's':
if (NextArg() != STRINGARG)
HError(1319,"HLStats: ENTER label name expected");
enterId=GetLabId(GetStrArg(),TRUE);
if (NextArg() != STRINGARG)
HError(1319,"HLStats: EXIT label name expected");
exitId=GetLabId(GetStrArg(),TRUE);
break;
case 't':
bigThresh = GetChkedInt(0,100,s);
break;
case 'u':
uniFloor = GetChkedFlt(0.0,1000.0,s);
break;
case 'G':
if (NextArg() != STRINGARG)
HError(1319,"HLStats: Input label File format expected");
if((ff = Str2Format(GetStrArg())) == ALIEN)
HError(-1389,"HLStats: Warning ALIEN Label file format set");
break;
case 'I':
if (NextArg() != STRINGARG)
HError(1319,"HLStats: Input MLF file name expected");
LoadMasterFile(GetStrArg());
break;
case 'T':
if (NextArg() != INTARG)
HError(1319,"HLStats: Trace value expected");
trace = GetChkedInt(0,017,s); break;
default:
HError(1319,"HLStats: Unknown switch %s",s);
}
}
if (NextArg()!=STRINGARG)
HError(1319,"HLStats: Label list file name expected");
listfn = GetStrArg();
if (!(doDurs || doBigram || doList || doLCount || doPCount))
HError(1330,"HLStats: Nothing to do!");
InitStats(listfn);
i=0;
while (NumArgs()>0) {
if (NextArg()!=STRINGARG)
HError(1319,"HLStats: Input label file name expected");
labFn = GetStrArg();
if (IsMLFFile(labFn)) {
fidx = NumMLFFiles();
if ((me=GetMLFTable()) != NULL) {
while(me->next != NULL) me=me->next;
LoadMasterFile(labFn);
me=me->next;
}
else {
LoadMasterFile(labFn);
me=GetMLFTable();
}
while (me != NULL) {
if (me->type == MLF_IMMEDIATE && me->def.immed.fidx == fidx) {
if (trace&T_FIL) {
printf(" Processing file %s\n",me->pattern); fflush(stdout);
}
t = LOpen(&tmpHeap,me->pattern,ff);
if (t->numLists<1)
HError(-1330,"HLStats: Empty file %s",me->pattern);
else
GatherStats(t),i++;
Dispose(&tmpHeap,t);
}
me = me->next;
if ((trace&T_BAS) && !(trace&T_FIL) &&
NumMLFEntries()>5000 && i%1000==0)
printf(". "),fflush(stdout);
}
if ((trace&T_BAS) && !(trace&T_FIL) && NumMLFEntries()>5000)
printf("\n");
} else {
if (trace&T_FIL) {
printf(" Processing file %s\n",labFn); fflush(stdout);
}
t = LOpen(&tmpHeap,labFn,ff);
if (t->numLists<1)
HError(-1330,"HLStats: Empty file %s",me->pattern);
else
GatherStats(t),i++;
Dispose(&tmpHeap,t);
}
}
if (trace&T_MEM)
PrintAllHeapStats();
OutputStats();
if (trace&T_MEM)
PrintAllHeapStats();
Exit(0);
return (0); /* never reached -- make compiler happy */
}
int main(int argc, char *argv[])
{
char *s;
void Initialise(void);
void DoRecognition(void);
void DoAlignment(void);
if(InitShell(argc,argv,hvite_version,hvite_vc_id)<SUCCESS)
HError(3200,"HVite: InitShell failed");
InitMem(); InitLabel();
InitMath(); InitSigP();
InitWave(); InitAudio();
InitVQ(); InitModel();
if(InitParm()<SUCCESS)
HError(3200,"HVite: InitParm failed");
InitDict();
InitNet(); InitRec();
InitUtil();
InitAdapt(&xfInfo); InitMap();
if (!InfoPrinted() && NumArgs() == 0)
ReportUsage();
if (NumArgs() == 0) Exit(0);
SetConfParms();
CreateHeap(&modelHeap, "Model heap", MSTAK, 1, 0.0, 100000, 800000 );
CreateHMMSet(&hset,&modelHeap,TRUE);
while (NextArg() == SWITCHARG) {
s = GetSwtArg();
if (strlen(s)!=1)
HError(3219,"HVite: Bad switch %s; must be single letter",s);
switch(s[0]){
case 'a':
loadLabels=TRUE; break;
case 'b':
if (NextArg()!=STRINGARG)
HError(3219,"HVite: Utterance boundary word expected");
bndId = GetLabId(GetStrArg(),TRUE); break;
case 'c':
tmBeam = GetChkedFlt(0.0,1000.0,s); break;
case 'd':
if (NextArg()!=STRINGARG)
HError(3219,"HVite: HMM definition directory expected");
hmmDir = GetStrArg(); break;
case 'e':
saveAudioOut=TRUE; break;
case 'f':
states=TRUE; break;
case 'g':
replay=TRUE; break;
case 'i':
if (NextArg()!=STRINGARG)
HError(3219,"HVite: Output MLF file name expected");
/* if(SaveToMasterfile(GetStrArg())<SUCCESS)
HError(3214,"HCopy: Cannot write to MLF"); */
SaveToMasterfile(GetStrArg());
break;
case 'k':
xfInfo.useInXForm = TRUE;
break;
case 'j':
if (NextArg()!=INTARG)
HError(3219,"HVite: No. of files per online adaptation step expected");
update = GetChkedInt(1,256,s);
break;
case 'l':
if (NextArg()!=STRINGARG)
HError(3219,"HVite: Label file directory expected");
labDir = GetStrArg(); break;
case 'm':
models=TRUE; break;
case 'n':
nToks = GetChkedInt(2,MAX_TOKS,s);
if (NextArg()==FLOATARG || NextArg()==INTARG)
nTrans = GetChkedInt(1,10000,s);
else
nTrans = 1;
break;
case 'o':
if (NextArg()!=STRINGARG)
HError(3219,"HVite: Output label format expected");
labForm = GetStrArg(); break;
case 'p':
wordPen = GetChkedFlt(-1000.0,1000.0,s); break;
case 'q':
if (NextArg()!=STRINGARG)
HError(3219,"HVite: Output lattice format expected");
latForm = GetStrArg(); break;
case 'r':
prScale = GetChkedFlt(0.0,1000.0,s); break;
case 's':
lmScale = GetChkedFlt(0.0,1000.0,s); break;
case 't':
genBeam = GetChkedFlt(0,1.0E20,s);
if (genBeam == 0.0)
genBeam = -LZERO;
if (NextArg()==FLOATARG || NextArg()==INTARG) {
genBeamInc = GetChkedFlt(0.0,1.0E20,s);
genBeamLim = GetChkedFlt(0.0,1.0E20,s);
if (genBeamLim < (genBeam + genBeamInc)) {
genBeamLim = genBeam; genBeamInc = 0.0;
}
}
else {
genBeamInc = 0.0;
genBeamLim = genBeam;
}
break;
case 'w':
if (NextArg()!=STRINGARG)
loadNetworks=TRUE;
else {
wdNetFn = GetStrArg();
if (strlen(wdNetFn)==0) {
wdNetFn=NULL;
loadNetworks=TRUE;
}
}
break;
case 'u':
maxActive = GetChkedInt(0,100000,s); break;
case 'v':
wordBeam = GetChkedFlt(0,1.0E20,s);
if (wordBeam == 0.0)
wordBeam = -LZERO;
break;
case 'x':
if (NextArg()!=STRINGARG)
HError(3219,"HVite: HMM file extension expected");
hmmExt = GetStrArg(); break;
case 'y':
if (NextArg()!=STRINGARG)
HError(3219,"HVite: Output label file extension expected");
labExt = GetStrArg(); break;
case 'z':
if (NextArg()!=STRINGARG)
HError(3219,"HVite: Lattice output file extension expected");
latExt = GetStrArg(); break;
case 'F':
if (NextArg() != STRINGARG)
HError(3219,"HVite: Data File format expected");
if((dfmt = Str2Format(GetStrArg())) == ALIEN)
HError(-3289,"HVite: Warning ALIEN Input file format set");
break;
case 'G':
if (NextArg() != STRINGARG)
HError(3219,"HVite: Source Label File format expected");
if((ifmt = Str2Format(GetStrArg())) == ALIEN)
HError(-3289,"HVite: Warning ALIEN Input file format set");
break;
case 'H':
if (NextArg() != STRINGARG)
HError(3219,"HVite: MMF File name expected");
AddMMF(&hset,GetStrArg());
break;
case 'I':
if (NextArg() != STRINGARG)
HError(3219,"HVite: MLF file name expected");
LoadMasterFile(GetStrArg()); break;
case 'L':
if (NextArg()!=STRINGARG)
HError(3219,"HVite: Label/network file directory expected");
labInDir = GetStrArg(); break;
case 'P':
if (NextArg() != STRINGARG)
HError(3219,"HVite: Target Label File format expected");
if((ofmt = Str2Format(GetStrArg())) == ALIEN)
HError(-3289,"HVite: Warning ALIEN Label output file format set");
break;
case 'B':
saveBinary = TRUE;
break;
case 'T':
trace = GetChkedInt(0,511,s); break;
case 'X':
if (NextArg()!=STRINGARG)
HError(3219,"HVite: Input label/network file extension expected");
labInExt = GetStrArg(); break;
case 'h':
if (NextArg()!=STRINGARG)
HError(1,"Speaker name pattern expected");
xfInfo.outSpkrPat = GetStrArg();
if (NextArg()==STRINGARG) {
xfInfo.inSpkrPat = GetStrArg();
if (NextArg()==STRINGARG)
xfInfo.paSpkrPat = GetStrArg();
}
if (NextArg() != SWITCHARG)
HError(2319,"HERest: cannot have -h as the last option");
break;
case 'E':
if (NextArg()!=STRINGARG)
HError(2319,"HERest: parent transform directory expected");
xfInfo.usePaXForm = TRUE;
xfInfo.paXFormDir = GetStrArg();
if (NextArg()==STRINGARG)
xfInfo.paXFormExt = GetStrArg();
if (NextArg() != SWITCHARG)
HError(2319,"HVite: cannot have -E as the last option");
break;
case 'J':
if (NextArg()!=STRINGARG)
HError(2319,"HERest: input transform directory expected");
AddInXFormDir(&hset,GetStrArg());
if (NextArg()==STRINGARG)
xfInfo.inXFormExt = GetStrArg();
if (NextArg() != SWITCHARG)
HError(2319,"HVite: cannot have -J as the last option");
break;
case 'K':
if (NextArg()!=STRINGARG)
HError(2319,"HVite: output transform directory expected");
xfInfo.outXFormDir = GetStrArg();
xfInfo.useOutXForm = TRUE;
if (NextArg()==STRINGARG)
xfInfo.outXFormExt = GetStrArg();
if (NextArg() != SWITCHARG)
HError(2319,"HVite: cannot have -K as the last option");
break;
default:
HError(3219,"HVite: Unknown switch %s",s);
}
}
if (NextArg()!=STRINGARG)
HError(3219,"HVite: Dictionary file name expected");
dictFn = GetStrArg();
if (NextArg()!=STRINGARG)
HError(3219,"HVite: HMM list file name expected");
hmmListFn = GetStrArg();
#ifndef PHNALG
if ((states || models) && nToks>1)
HError(3230,"HVite: Alignment using multiple tokens is not supported");
#endif
if (NumArgs()==0 && wdNetFn==NULL)
HError(3230,"HVite: Network must be specified for recognition from audio");
if (loadNetworks && loadLabels)
HError(3230,"HVite: Must choose either alignment from network or labels");
if (nToks>1 && latExt==NULL && nTrans==1)
HError(-3230,"HVite: Performing nbest recognition with no nbest output");
if (nToks > 1 && latExt != NULL && nTrans > 1)
HError(-3230,"HVite: Performing nbest recognition with 1-best and latttices output");
if ((update>0) && (!xfInfo.useOutXForm))
HError(3230,"HVite: Must use -K option with incremental adaptation");
Initialise();
/* Process the data */
if (wdNetFn==NULL)
DoAlignment();
else
DoRecognition();
/* Free up and we are done */
if (trace & T_MEM) {
printf("Memory State on Completion\n");
PrintAllHeapStats();
}
DeleteVRecInfo(vri);
ResetHeap(&netHeap);
FreePSetInfo(psi);
UpdateSpkrStats(&hset,&xfInfo, NULL);
ResetHeap(®Heap);
ResetHeap(&modelHeap);
Exit(0);
return (0); /* never reached -- make compiler happy */
}