hlabel.c

来自「隐马尔科夫模型工具箱」· C语言 代码 · 共 1,565 行 · 第 1/3 页

   printf("\n");
}

/* PrintList: print the given label list */
static void PrintList(LabList *ll)
{
   int i;
   LLink p;
   
   for (p=ll->head->succ,i=1; p->succ!= NULL; p=p->succ,i++) {
      printf("%4d. ",i);
      PrintLabel(p,ll->maxAuxLab);
   }
}

/* EXPORT->PrintTranscription: for diagnostics */
void PrintTranscription(Transcription *t, char *title)
{
   int i;
   LabList *ll;
   
   printf("Transcription: %s [%d lists]\n",title,t->numLists);
   ll = t->head;
   for (i=1; i<=t->numLists; i++) {
      printf(" List %d\n",i);
      PrintList(ll); ll = ll->next;
   }
}

/* FilterLevel: remove all but given level from transcription */
static void FilterLevel(Transcription *t, int lev)
{
   LabList *ll;
   LLink p;   
   
   for (ll = t->head; ll != NULL; ll = ll->next) {
      if (ll->maxAuxLab < lev)
         HError(6570,"FilterLevel: level %d > max[%d]",lev,ll->maxAuxLab);
      if (lev > 0) {
         for (p=ll->head->succ; p->succ!=NULL; p=p->succ) {
            if (p->auxLab[lev] != NULL) {
               p->labid = p->auxLab[lev];
               p->score = p->auxScore[lev];
               p->end = AuxLabEndTime(p,lev);
            } else
               DeleteLabel(p);
         }
      }
      ll->maxAuxLab = 0;
   }
}

/* --------------- Format Specific Label File Input Routines --------------- */

#define LEVELSEP "///"
#define strlen_LEVELSEP 3
#define COMMCHAR  ';'
#define LFEED   '\012'
#define CRETURN '\015'

enum _TrSymbol{TRNULL,TRNUM,TRSTR,TREOL,TRLEV,TRCOMMA,TREOF};
typedef enum _TrSymbol TrSymbol;

static int curch = ' ';
static TrSymbol trSym = TRNULL;
static double trNum;
static char trStr[256];

/* IsNumeric: returns true if given string is a number */
Boolean IsNumeric(char *s)
{
   int i,len;
   
#ifdef WIN32
   if(*s < 0)
      return FALSE;
#endif

   len = strlen(s)-1;
   if (!(isdigit(s[0])||s[0]=='+'||s[0]=='-')) return FALSE;
   if (!(isdigit(s[len]))) return FALSE;   
   for (i=1; i<len; i++)
      if (!(isdigit(s[i]) || s[i]=='.' || s[i]=='-' || s[i]=='+' || s[i]=='e' || s[i]=='E' ))
         return FALSE;
   return TRUE;
}

/* InitTrScan: initialise the scanner */
static void InitTrScan(void)
{
   curch = ' ';
   trSym = TRNULL;
}

/* GetTrSym: get next symbol from f, remember that f might be an MLF
             in which case EOF is a period on its own line   */
static void GetTrSym(Source *src, Boolean htk)
{
   int nxtch;
   Boolean trSOL;

   trNum = 0.0; trStr[0]='\0'; 
   if (trSym==TRNULL) curch = GetCh(src);
   if (trSym==TREOL || trSym==TRNULL)
      trSOL=TRUE;
   else
      trSOL=FALSE;
   while (curch == ' ' || curch == '\t') {
      trSOL=FALSE;
      curch = GetCh(src);
   }
   if (!htk && curch == COMMCHAR)
      SkipLine(src);
   
   switch (curch) {
   case EOF:
      trSym = TREOF;
      break;
   case LFEED:
      curch = GetCh(src);
      trSym = TREOL;
      break;
   case CRETURN:
      curch = GetCh(src);
      if (curch == LFEED) curch = GetCh(src);
      trSym = TREOL;
      break;
   case ',':
      if (!htk) {
         curch = GetCh(src); trSym = TRCOMMA;
         break;
      }
   case '.':
      if (curch=='.' && trSOL==TRUE && mlfUsed>0 && htk) {
         nxtch = GetCh(src);
         if (nxtch == LFEED  || nxtch == CRETURN) {
            trSym = TREOF;
            break;
         }
         UnGetCh(nxtch,src);  /*  Requires more than one character pushback */
      }
   default:
      if (htk) {
         UnGetCh(curch,src);
         if (!ReadString(src,trStr)) {
            trSym=TREOF;
            break;
         }
         curch=GetCh(src);
         if (trSOL && strcmp(LEVELSEP,trStr)==0) {
            if (curch == LFEED  || curch == CRETURN) {
               trSym = TRLEV;
               break;
            }
         }
      }
      else {
         nxtch=0;
         do {
            if (nxtch>=255) break;
            trStr[nxtch++]=curch; curch=GetCh(src);
         }
         while (!isspace(curch) && curch != ',' && curch != EOF);
         trStr[nxtch]='\0';
         src->wasQuoted=FALSE;
      }
      if (!src->wasQuoted && IsNumeric(trStr)){
         sscanf(trStr,"%lf",&trNum);
         trSym = TRNUM;
         break;
      }
      if (htk && compatMode && 
          (strcmp(LEVELSEP,trStr)==0 || strcmp(".",trStr)==0)) {
         src->wasNewline=FALSE;
         SkipWhiteSpace(src);
         if (src->wasNewline) {
            curch = CRETURN;
            trSym=(strcmp(LEVELSEP,trStr)==0?TRLEV:TREOF);
            break;
         }
         curch = GetCh(src);
      }
      if (stripTriPhones) TriStrip(trStr);
      trSym = TRSTR;
      break;
   }
}

/* -------------------- HTK Label Format ------------------ */

/* ExtendAux: extend the aux arrays in given lab list to n elems 
   New elems are set to NULL/0.0 */
static void ExtendAux(MemHeap *x, LabList *ll, int n)
{
   int i,oldn;
   LabId *id;
   float *s;
   LLink p;

   if (n>=99)
      HError(6570, "ExtendAux: Too many auxiliary fields in label file");
   
   oldn = ll->maxAuxLab; ll->maxAuxLab = n;
   for (p=ll->head->succ; p->succ!=NULL; p=p->succ){
      id = (LabId *)New(x,sizeof(LabId)*n) - 1; 
      s = (float *)New(x,sizeof(float)*n) - 1;
      for (i=1; i<=oldn; i++){
         id[i] = p->auxLab[i];
         s[i] = p->auxScore[i];
      }
      for (i=oldn+1; i<=n; i++){
         id[i] = NULL;
         s[i] = 0.0;
      }
      p->auxLab = id; p->auxScore = s;
   }
   if (trace&T_HTKL)
      printf("HLabel:     aux extended from %d to %d\n",oldn,n);
}

/* LoadHTKList: load a single HTK label list - dont create anything if 
                transAlt>0 and alt != transAlt */
static LabList * LoadHTKList(MemHeap *x, Source *src, int alt)
{
   LabList  *ll = NULL;
   LabId labid, auxLab[100];
   LLink p;
   HTime start,end;
   float score, auxScore[100];
   int n,maxAux = 0;
   Boolean ok;
   
   ok = (transAlt==0) || (transAlt == alt);
   if (ok) ll = CreateLabelList(x,maxAux);  /* assume no aux labels */
   if (trace&T_HTKL)
      printf("HLabel: looking for lab list\n");
   
   while (trSym==TRNUM || trSym==TRSTR){
      start = -1; end = -1; score = 0.0;
      if (trSym==TRNUM) {
         start = trNum; GetTrSym(src,TRUE);
         if (trSym==TRNUM) {
            end = trNum; GetTrSym(src,TRUE);
         }
      }
      if (trSym != TRSTR)
         HError(6550,"LoadHTKList: Label Name Expected");
      labid = GetLabId(trStr,TRUE);
      GetTrSym(src,TRUE);
      if (trSym==TRNUM){
         score = trNum;
         GetTrSym(src,TRUE);
      }
      if (trace&T_HTKL)
         printf("HLabel: adding %.0f %.0f %s %f\n",start,end,labid->name,score);
      if (ok) p = AddLabel(x,ll,labid,start,end,score);
      /* Any aux labels ? */
      n = 0;
      while (trSym != TREOL && trSym!=TREOF) {
         n++;
         if (trSym != TRSTR)
            HError(6550,"LoadHTKList: Aux Label Name Expected");
         auxLab[n] = GetLabId(trStr,TRUE);
         if (trace&T_HTKL)
            printf("HLabel:   adding aux lab %d = %s\n",n,auxLab[n]->name);
         GetTrSym(src,TRUE);
         if (trSym==TRNUM){
            auxScore[n] = trNum;
            if (trace&T_HTKL)
               printf("HLabel:   adding aux score %d = %f\n",n,trNum);
            GetTrSym(src,TRUE);
         } else
            auxScore[n] = 0.0;
      }
      if (ok && n>0) { /* need to add aux info */
         if (n>maxAux) {
            ExtendAux(x,ll,n);
            maxAux = n;
         } else while (n<maxAux) {
            ++n;
            auxLab[n] = NULL;
            auxScore[n] = 0.0;
         }
         AddAuxLab(p,n,auxLab,auxScore);
      }
      if (trSym!=TREOF)
         GetTrSym(src,TRUE);
   }
   return ll;
}

/* LoadHTKLabels: load a HTK transcription */
static void LoadHTKLabels(MemHeap *x, Transcription *t, Source *src)
{
   LabList *ll;
   int alt = 0;
   
   InitTrScan();
   GetTrSym(src,TRUE);
   if (trSym==TRNUM || trSym==TRSTR){
      ll = LoadHTKList(x,src,++alt);
      AddLabelList(ll,t);
      while (trSym == TRLEV){
         GetTrSym(src,TRUE);
         if (trSym != TREOL)
            HError(6550,"LoadHTKList: End of Line after /// Expected");       
         GetTrSym(src,TRUE);
         ll = LoadHTKList(x,src,++alt);
         AddLabelList(ll,t);
      }
   }
   while (trSym==TREOL) 
      GetTrSym(src,TRUE);
   if (trSym != TREOF)
      HError(6550,"LoadHTKLabels: Junk at end of HTK transcription");
}

/* --------------- TIMIT Label Format --------------------- */

/* LoadTIMITLabels: load a TIMIT transcription */
static void LoadTIMITLabels(MemHeap *x, Transcription *t, Source *src)
{
   LabList *ll;
   LabId labid;
   HTime start,end;
   float score;
   
   ll = CreateLabelList(x,0);  AddLabelList(ll,t);
   InitTrScan();
   GetTrSym(src,FALSE);
   while (trSym == TRNUM){    
      start = trNum*625;                 /* sample rate is 16KHz */
      GetTrSym(src,FALSE);
      if (trSym != TRNUM) 
         HError(6552,"LoadTIMITLabels: End Time expected in TIMIT Label File");
      end   = trNum*625;
      GetTrSym(src,FALSE);
      if (trSym != TRSTR) 
         HError(6552,"LoadTIMITLabels: Label Name expected in TIMIT Label File");
      labid = GetLabId(trStr,TRUE);
      score = 0.0;
      AddLabel(x,ll,labid,start,end,score);
      GetTrSym(src,FALSE);
      if (trSym == TREOL)
         GetTrSym(src,FALSE);
   }
}

/* --------------- ESPS Label Format --------------------- */

/* LoadESPSLabels: read waves label file */
static void LoadESPSLabels(MemHeap *x, Transcription *t, Source *src)
{
   LabList *ll;
   LabId labid;
   HTime start,end;
   float score;
   
   ll = CreateLabelList(x,0);  AddLabelList(ll,t);
   InitTrScan();
   GetTrSym(src,FALSE);
   while ( trStr[0] != '#' ) {
      GetTrSym(src,FALSE);
      if (trSym == TREOF) 
         HError(6553,"LoadESPSLabels: Unexpected EOF in ESPS Waves Label file.");
      if ( strcmp( "nfields", trStr) == 0 ) {
         GetTrSym(src,FALSE);
         if ( trSym != TRNUM )
            HError(6553,"LoadESPSLabels: Expecting field number");
         if ( trNum != 1 )
            HError(6553,"LoadESPSLabels: Can only read single field label files.");
      }
   }
   end = start = 0.0; score = 0.0;
   GetTrSym(src,FALSE);
   while ( trSym != TRNUM && trSym != TREOF )
      GetTrSym(src,FALSE);
   while ( trSym == TRNUM ) {    
      start = end;   /* Get time stamp */
      end = trNum * 1.0E7; 
      if ( start > end )
         HError(-6553,"LoadESPSLabels: time stamps out of order.");
      GetTrSym(src,FALSE);       /* Ignore color */
      GetTrSym(src,FALSE);       /* Get field label for current level */
      if ( trSym != TRSTR) 
         HError(6553,"LoadESPSLabels: Expecting label string in ESPS Waves Label file.");
      labid = GetLabId( trStr,TRUE); 
      AddLabel(x,ll,labid,start,end,score);  
      GetTrSym(src,FALSE);
      if ( trSym != TREOL )
         HError(6553,"LoadESPSLabels: End-of-line expected in ESPS Waves Label file.");
      GetTrSym(src,FALSE);
   }
}

/* --------------- SCRIBE Label Format --------------------- */

enum _ScribeLab {
   S_LBB, S_LBA, S_UTS, S_EOF
};
typedef enum _ScribeLab ScribeLab;

static char *scribeMap[] = {"LBB:", "LBA:", "UTS:"};

/* GetScribeLab: get next scribe label in input */
static ScribeLab GetScribeLab(Source *src)
{
   char *s;
   int i;
   
   do {
      do {
         GetTrSym(src,FALSE);
         if (trSym == TREOF) return S_EOF;
      } while (trSym != TRSTR);
      i = 0;
      do {
         s=scribeMap[i]; 
         if (strcmp(trStr,s) == 0) 
            return (ScribeLab) i;
         i++;
      } while (i < S_EOF);
   } while(TRUE);
   return ((ScribeLab) 0);
}

/* LoadSCRIBELabels: load a SCRIBE (SAM) label file - searches for
         first occurrence of a label symbol 
               LBA - acoustic label
               LBB - broad class label
               UTS - utterance 
         it loads this symbol and all subsequent labels of the
         same type.  All other SAM label types are ignored */
static void LoadSCRIBELabels(MemHeap *x, Transcription *t, Source *src)
{
   LabList *ll;
   LabId labid;
   HTime start,end;
   float score;
   ScribeLab ltype, lx;
   double sp;
   char buf[256];
   
   if (!GetConfFlt(cParm,numParm,"SOURCERATE",&sp))  
      sp = 500.0;   /* actual SCRIBE rate */
   ll = CreateLabelList(x,0);  AddLabelList(ll,t);
   InitTrScan();
   do {  /* search for first label */
      ltype = GetScribeLab(src);
      if (ltype == S_EOF)
         HError(6554,"LoadSCRIBELabels: Unexpected EOF");
   } while (ltype != S_LBB && ltype != S_LBA && ltype != S_UTS);
   do { /* load this and all subsequent ltype labels */
      GetTrSym(src,FALSE);
      if (trSym != TRNUM)
         HError(6554,"LoadSCRIBELabels: Start Index expected [%d]\n",trSym);
      start = trNum * sp;
      GetTrSym(src,FALSE);
      if (trSym != TRCOMMA)
         HError(6554,"LoadSCRIBELabels: Comma expected [%d]\n",trSym);
      GetTrSym(src,FALSE);
      if (ltype == S_LBA || ltype == S_LBB) {   /* LBB and LBA have a centre field */
         if (trSym != TRCOMMA)
            HError(6554,"LoadSCRIBELabels: Comma expected [%d]\n",trSym);     
         GetTrSym(src,FALSE);
      }
      if (trSym != TRNUM)
         HError(6554,"LoadSCRIBELabels: End Index expected [%d]\n",trSym);
      end = trNum * sp;
      GetTrSym(src,FALSE);
      if (trSym != TRCOMMA)
         HError(6554,"LoadSCRIBELabels: Comma expected [%d]\n",trSym);
      GetTrSym(src,FALSE);
      if (trSym != TRSTR)
         HError(6554,"LoadSCRIBELabels: Label expected [%d]\n",trSym);
      strcpy(buf,trStr);
      GetTrSym(src,FALSE);
      while (trSym == TRSTR){
         strcat(buf,"_"); strcat(buf,trStr);
         GetTrSym(src,FALSE);
      }
      labid = GetLabId(buf,TRUE); 
      score = 0.0;
      AddLabel(x,ll,labid,start,end,score); 
      if (trSym != TREOL)
         HError(6554,"LoadSCRIBELabels: End of Line expected [%d]\n",trSym);
      lx = GetScribeLab(src);
   } while (lx != S_EOF);
}

/* ----------------- TriPhone Stripping ------------------- */

/* EXPORT->TriStrip: Remove contexts of form A- and +B from s */
void TriStrip(char *s)
{
   char buf[100],*p;
   
   if ((p = strchr(s,'-')) == NULL) p = s; else ++p;
   strcpy(buf,p);
   if ((p = strrchr(buf,'+')) != NULL) 
      *p = '\0';
   strcpy(s,buf);
}

/* EXPORT->LTriStrip: enable triphone stripping */
void LTriStrip(Boolean enab)
{
   stripTriPhones = enab;
}

/* ------------------ Master Label File Handling -------------------- */

/* StoreMLFEntry: store the given MLF entry */
static void StoreMLFEntry(MLFEntry *e)
{
   e->next = NULL;
   if (mlfHead == NULL)
      mlfHead = mlfTail = e;
   else {
      mlfTail->next = e; mlfTail = e;
   }
   ++mlfUsed;