hlabel.c

实现HMM算法

/* EXPORT->NumCases: find num cases of primary label in label list */
int NumCases(LabList *ll, LabId id)
{
   int n = 0;
   LLink p;
   
   for (p=ll->head->succ; p->succ!=NULL; p=p->succ) 
      if (p->labid == id) ++n;
   return n;
}

/* EXPORT->GetCase: find nth occ of primary label in label list */
LLink GetCase(LabList *ll, LabId id, int n)
{
   LLink p;
   int k=0;
   
   for (p=ll->head->succ; p->succ!=NULL; p=p->succ) {
      if (p->labid == id) 
         if (++k==n) return p;
   }
   HError(6571,"GetCase: %d case of %s nonexistent",n,id->name);
   return NULL;
}

/* EXPORT->NumAuxCases: find num cases of aux label i in label list */
int NumAuxCases(LabList *ll, LabId id, int i)
{
   int n = 0;
   LLink p;
   
   if (ll->maxAuxLab < i)
      HError(6570,"NumAuxCases: aux idx %d > max[%d]",i,ll->maxAuxLab);
   for (p=ll->head->succ; p->succ!=NULL; p=p->succ) 
      if ((i==0) ? (p->labid==id) : (p->auxLab[i]==id))  ++n;
   return n;
}

/* EXPORT->GetAuxCase: find nth occ of aux label i in label list */
LLink GetAuxCase(LabList *ll, LabId id, int n, int i)
{
   LLink p;
   int k=0;
   
   if (ll->maxAuxLab < i)
      HError(6570,"GetAuxCase: aux idx %d > max[%d]",i,ll->maxAuxLab);
   for (p=ll->head->succ; p->succ!=NULL; p=p->succ) {
      if ((i==0) ? (p->labid==id) : (p->auxLab[i]==id))
         if (++k==n) return p;
   }
   HError(6571,"GetAuxCase: %d case of %s nonexistent",n,id->name);
   return NULL;
}

/* EXPORT->GetLabN: return n'th primary label */
LLink GetLabN(LabList *ll, int n)
{
   int count=0;
   LLink p;

   for (p=ll->head->succ; p->succ!= NULL; p=p->succ) 
      if (++count==n) return p;
   HError(6571,"GetLabN: %d'th label nonexistent",n);
   return NULL;
}

/* EXPORT->GetAuxLabN: return n'th aux i label */
LLink GetAuxLabN(LabList *ll, int n, int i)
{
   int count=0;
   LLink p;

   if (ll->maxAuxLab < i)
      HError(6570,"GetAuxLabN: aux idx %d > max[%d]",i,ll->maxAuxLab);
   for (p=ll->head->succ; p->succ!= NULL; p=p->succ) 
      if (i==0 || p->auxLab[i]!=NULL)      
         if (++count==n) return p;
   HError(6571,"GetAuxLabN: %d'th aux[%d] label nonexistent",n,i);
   return NULL;
}

/* EXPORT->CountLabs: count num primary labels in lab list */
int CountLabs(LabList *ll)
{
   int count = 0;
   LLink p;

   if (ll!=NULL)
      for (p=ll->head->succ; p->succ!= NULL; p=p->succ) 
         ++count;
   return(count);
}

/* EXPORT->CountAuxLabs: count num aux i labels in lab list */
int CountAuxLabs(LabList *ll, int i)
{
   int count = 0;
   LLink p;

   if (ll!=NULL)
      for (p=ll->head->succ; p->succ != NULL; p=p->succ)
         if (i==0 || p->auxLab[i]!=NULL)
            ++count;
   return(count);
}

/* EXPORT->AuxLabEndTime: return the end time for the i'th aux lab */
HTime AuxLabEndTime(LLink p, int i)
{
   LLink q;
   
   q = p->succ;
   while (i!=0 && q->succ != NULL && q->auxLab[i]==NULL){
      p = q; q = q->succ;
   }
   return p->end;
}

/* PrintLabel: print the given label on one line */
static void PrintLabel(LLink p, int maxAux)
{
   int n;
   LabId id;
   
   printf("%8.0f%8.0f",p->start,p->end);
   printf(" %8s %5f",p->labid->name,p->score);
   for (n=1; n<=maxAux; n++){
      id = p->auxLab[n];
      printf(" %8s %5f",(id!=NULL)?id->name:"<null>",p->auxScore[n]);
   }
   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((int) s[0])||s[0]=='+'||s[0]=='-')) return FALSE;
   if (!(isdigit((int) s[len]))) return FALSE;   
   for (i=1; i<len; i++)
      if (!(isdigit((int) 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) {/*当前htk为真，执行此分支*/
         UnGetCh(curch,src);
         if (!ReadString(src,trStr)) {
            trSym=TREOF;
            break;
         }/*读出的标签描述字符串保存在trStr中，如trStr=oo,即音素单元名称*/
         curch=GetCh(src);
         if (trSOL && strcmp(LEVELSEP,trStr)==0) {
            if (curch == LFEED  || curch == CRETURN) {
               trSym = TRLEV;
               break;
            }
         }
      }
      else {/*此分支不执行*/
         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;
   }/*switch语句到此结束*/
}

/* -------------------- 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 = NULL;
   HTime start,end;
   float score, auxScore[100];
   int n,maxAux = 0;
   Boolean ok;
   
   ok = (transAlt==0) || (transAlt == alt);/*当前ok为真*/
   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){/*当前trSym==TRSTR*/
      start = -1; end = -1; score = 0.0;
      if (trSym==TRNUM) {
         start = trNum; GetTrSym(src,TRUE);
         start *= htkLabelTimeScale;
         if (trSym==TRNUM) {
            end = trNum; GetTrSym(src,TRUE);
            end *= htkLabelTimeScale;
         }
      }
      if (trSym != TRSTR)
         HError(6550,"LoadHTKList: Label Name Expected");
      labid = GetLabId(trStr,TRUE);/*用描述标签字符串生成标签id，labid的name即为该标签字符串，如OO*/
      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;/*当前trSym=TREOL,故不执行下面的循环*/
      while (trSym != TREOL && trSym!=TREOF) {