hlabel.c

实现HMM算法

         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) {/*当前n=0，故不执行此分支*/ /* 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){/*当前不执行此循环*/
         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;
}

/* FindMLFStr: find the next quoted string in s */
static Boolean FindMLFStr(char *s, char **st, char **en)
{
   char *p,*q;
   
   if (s==NULL || *s == '\0') return FALSE;
   p = strchr(s,'"');//查找字符串s中首次出现字符""的位置
   if (p==NULL || *(p+1)=='\0') return FALSE;
   q = strchr(p+1,'"');
   if (q==NULL) return FALSE;
   *st = p; *en = q;
   return TRUE;
}

/* FindMLFType: find a -> or => symbol in s if any */
static MLFDefType FindMLFType(char *s, char **en)
{
   char *p;
   
   p=strchr(s+1,'>');
   if (p!=NULL) {
      *en = p;
      if (*(p-1) == '-') return MLF_SIMPLE;
      if (*(p-1) == '=') return MLF_FULL;
   }
   return MLF_IMMEDIATE;
}

/* NoMLFHeader: return true if s doesnt contain #!MLF!# */
static Boolean NoMLFHeader(char *s)
{
   int len;
   char *e;
   
   len = strlen(s);
   while (isspace((int) *s) && len>7) {
      --len; ++s;
   }
   e = s+len-1;
   while (isspace((int) *e) && len>7) {
      --len; --e;
   }
   if (len != 7) return TRUE;
   *(e+1) = '\0';
   return (strcmp(s,"#!MLF!#") != 0);
}

static Boolean incSpaces;

/* IsDotLine: return true if line contains only a single dot */
static Boolean IsDotLine(char *s)
{
   int len;
   char *e;
   Boolean cut;

   len = strlen(s);cut=FALSE;
   if (s[len-1]=='\n') {
      len--; s[len]=0;
      if ((len>0) && (s[len-1]=='\r')) {
         len--; s[len]=0;
      }
   }
   if (compatMode) {
      while (isspace((int) *s) && len>0) {
         cut=TRUE;
         --len; ++s;
      }
      e = s+len-1;
      if (*e=='\n' && len>0) --len,--e;
      while (isspace((int) *e) && len>0) {
         cut=TRUE;
         --len; --e;
      }
   }
   if ((*s == '.') && (len == 1)) {
      if (cut) incSpaces=TRUE;
      return(TRUE);
   }
   if (compatMode && cut && len==strlen_LEVELSEP && 
       strncmp(s,LEVELSEP,strlen_LEVELSEP)==0)
      incSpaces=TRUE;
   return(FALSE);
}

/* ClassifyMLFPattern: classify given pattern string */
MLFPatType ClassifyMLFPattern(char *s)
{
   char *t;
   
   if (strchr(s,'?') != NULL) return PAT_GENERAL;
   if (strchr(s,'*') == NULL) return PAT_FIXED;
   if (strlen(s)<=2) return PAT_GENERAL;
   if (s[0]!='*' || s[1]!=PATHCHAR) return PAT_GENERAL;
   t = s+2;
   if (strchr(t,'*') == NULL ) return PAT_ANYPATH;
   return PAT_GENERAL;
}

/* MLFHash: hash the given string */
static unsigned MLFHash(char *s)
{
   unsigned hashval;

   for (hashval=0; *s != '\0'; s++)
      hashval = *s + 31*hashval;
   return hashval;
}

/* EXPORT->LoadMasterFile: Load the Master Label File stored in fname 
                           and append the entries to the MLF table */
void LoadMasterFile(char *fname)
{
   char buf[1024];
   char *men;        /* end of mode indicator */
   char *pst,*pen;   /* start/end of pattern (inc quotes) */
   char *dst,*den;   /* start/end of subdirectory (inc quotes) */
   Boolean inEntry = FALSE;   /* ignore ".." within an entry */
   MLFEntry *e;
   FILE *f;
   
   if (numMLFs == MAXMLFS)
      HError(6520,"LoadMasterFile: MLF file limit reached [%d]",MAXMLFS);
   if ((f = fopen(fname,"rb")) == NULL)
      HError(6510,"LoadMasterFile: cannot open MLF %s",fname);
   if (fgets(buf,1024,f) == NULL)
      HError(6513,"LoadMasterFile: MLF file is empty");
   if (NoMLFHeader(buf))