hled.c

该压缩包为最新版htk的源代码,htk是现在比较流行的语音处理软件,请有兴趣的朋友下载使用

/* ----------------------------------------------------------- */
/*                                                             */
/*                          ___                                */
/*                       |_| | |_/   SPEECH                    */
/*                       | | | | \   RECOGNITION               */
/*                       =========   SOFTWARE                  */ 
/*                                                             */
/*                                                             */
/* ----------------------------------------------------------- */
/*         Copyright: Microsoft Corporation                    */
/*          1995-2000 Redmond, Washington USA                  */
/*                    http://www.microsoft.com                 */
/*                                                             */
/*   Use of this software is governed by a License Agreement   */
/*    ** See the file License for the Conditions of Use  **    */
/*    **     This banner notice must not be removed      **    */
/*                                                             */
/* ----------------------------------------------------------- */
/*      File: HLEd.c: Edit label file(s)                       */
/* ----------------------------------------------------------- */

char *hled_version = "!HVER!HLEd:   3.3 [CUED 28/04/05]";
char *hled_vc_id = "$Id: HLEd.c,v 1.2 2005/05/12 15:51:28 jal58 Exp $";

#include "HShell.h"
#include "HMem.h"
#include "HMath.h"
#include "HWave.h"
#include "HLabel.h"
#include "HDict.h"


/*
   This program reads in a list of 'editing' commands from a
   script file and then makes an edited copy of one or more 
   label files.  Each edit command in the script file must be
   on a separate line.  The commands supported are printed by
   the Summary routine.  Some commands can be 1 or 2 letters.
   The 1 letter form is retained from the original for compatibility.
*/

/* -------------------------- Trace Flags & Vars ------------------------ */

#define T_TOP     0001           /* basic progress reporting */
#define T_RDSC    0002           /* output edit script details */
#define T_CMDX    0004           /* trace general command operation */
#define T_CHOP    0010           /* trace Change operation */
#define T_SLEV    0020           /* trace level split/merge operation */
#define T_DLEV    0040           /* trace delete level operation */
#define T_EDIN    0100           /* trace edit file input */
#define T_MEM     0200           /* trace memory usage */
#define T_DIC     0400           /* trace dictionary substitution */

static int  trace    = 0;           /* trace level */
static ConfParam *cParm[MAXGLOBS];   /* configuration parameters */
static int nParm = 0;               /* total num params */

/* -------------------------- Global Variables etc ---------------------- */

#define MAXARGS  50     /* in any single HLEd command */
#define MAXIW    10     /* max number of interword boundaries */
#define MAXIV    10     /* max number of non-interword boundaries */

static char * newDir    = NULL;     /* dest label file directory */
static char * newExt    = "lab";    /* dest label file extension */

typedef enum { 
   NOCMD=0, 
   REPLACE, CHANGE, FIND,   MERGE,  EDOP_DELETE, DEFCON, TRIST,  
   SBDEF,   EXPAND, IFILL,  SORT,   WBDEF,  VBDEF,  LCTXT,  
   RCTXT,   TCTXT,  SETLEV, DELLEV, SPLLEV, ISIL,
   LASTCMD
} EdOp;

static int  nCmds = LASTCMD-1;
static char *cmdmap[] = {"","RE","CH","FI","ME","DE","DC","IT",
                         "SB","EX","FG","SO","WB","NB","LC",
                         "RC","TC","ML","DL","SP","IS","" };
static int  n1Cmds = 17;
static char oldmap[] = " RCFMDXZBEISWVLGTY ";

typedef struct {
   EdOp op;
   short nArgs;
   LabId args[MAXARGS];
}EditCmd;

typedef struct _ScriptItem{
   EditCmd cmd;
   struct _ScriptItem *next;
}ScriptItem;

static ScriptItem *script;          /* linked list of edit commands */
static LabId asterix;               /* id of an asterix */
static LabId sentMarker=NULL;       /* sentence bndary marker set by SB cmd */
static FILE *newLabs = NULL;        /* list of newly created labels */
static LabId wbnd[MAXIW];           /* interword bndaries set via WB cmd */
static LabId vbnd[MAXIV];           /* non-interwd bndaries set via NB cmd */
static int nWB = 0;                 /* number of interword boundaries */
static int nVB = 0;                 /* number of non-interword boundaries */
static Boolean noBounds = FALSE;    /* suppress boundary times if TRUE */
static Boolean triStrip = FALSE;    /* ignore triphone contexts in matching */
static Boolean levSplit = FALSE;    /* Split levels into lists */
static Boolean sortFirst = FALSE;   /* Sort on loading */
static FileFormat ifmt=UNDEFF;      /* Label input file format */
static FileFormat ofmt=UNDEFF;      /* Label output file format */
static HTime minGap = 50000.0;      /* minimum interword gap */
static char *dictFn = NULL;         /* Dictionary used for EX command */
static Vocab vocab;                 /* And the associated vocab */

static MemHeap tempHeap;            /* Storage for current file */
static MemHeap permHeap;            /* Permanent storage */

/* ------------------ Process Command Line ------------------------- */

/* Summary: print a summary of all HLEd commands */
void Summary(void)
{
   printf("\nHLEd Command Summary\n\n");
   printf("CH/C X A Y B    - replace Y in context of A_B by X\n");
   printf("DC/X A B C ...  - define context A as labels B,C,...\n");
   printf("DE/D A B  ...   - delete labels A,B,...\n");
   printf("DL [N]          - delete all current level [or level N]\n");
   printf("EX/E            - expand labels from dictionary or form A_B_C\n");
   printf("FG/I X          - fill interlabel gaps with label X\n");
   printf("FI/F A Y B      - find all occurrences of pattern AYB\n");
   printf("IS A B          - insert A at start and B at end\n");
   printf("IT/Z            - ignore triphone contexts in CH/FI cmds\n");
   printf("LC/L [X]        - convert phonemes to Left-context dependent\n");
   printf("ML N            - move to level N (1-99)\n");
   printf("ME/M X A B ...  - merge label seq A,B,.. and rename X\n");
   printf("NB/V X          - label X is not an interword boundary\n");
   printf("RC/G [X]        - convert phonemes to right-context dependent\n");
   printf("RE/R X A B ...  - replace labels A,B,... by X\n");
   printf("SB/B X          - define X as a sentence boundary marker\n");
   printf("SO/S            - sort labels into time order\n");
   printf("SP              - split multiple levels into multiple lists\n");
   printf("TC/T [X [Y]]    - convert phonemes to Triphones\n");
   printf("WB/W X          - define X as an interword boundary\n\n");
   Exit(0);
}

void ReportUsage(void)
{
   printf("\nUSAGE: HLEd [options] edCmdFile labFiles...\n\n");
   printf(" Option                                       Default\n\n");
   printf(" -b      suppress boundary times              off\n");
   printf(" -d s    load dictionary from s               off\n");
   printf(" -i s    Output transcriptions to MLF s       off\n"); 
   printf(" -g f    Set min gap for FG command to f      50000.0\n"); 
   printf(" -l s    Dir to store output label file(s)    current\n");
   printf(" -m      Strip to monophones on loading       off\n");
   printf(" -n f    Output list of all new labs to f     off\n");
   PrintStdOpts("GIPQX");
   printf("\n\n");
}

int main(int argc, char *argv[])
{
   char * labFn, *scriptFn, *newFn=NULL, *s;
   int fidx;
   MLFEntry *me;
   void Initialise(void);
   void EditFile(char *labfn);
   void ReadScript(char *scriptFn);
   void ListFinds(void);

   if(InitShell(argc,argv,hled_version,hled_vc_id)<SUCCESS)
      HError(1200,"HLEd: InitShell failed");

   InitMem();   InitMath();
   InitWave();  InitLabel();
   InitDict();

   if (!InfoPrinted() && NumArgs() == 0)
      ReportUsage();
   if (NumArgs() == 0) Exit(0);

   Initialise();
   while (NextArg() == SWITCHARG) {
      s = GetSwtArg();
      if (strlen(s)!=1) 
         HError(1219,"HLEd: Bad switch %s; must be single letter",s);
      switch(s[0]){
      case 'b':
         noBounds = TRUE; break;
      case 'd':
         if (NextArg()!=STRINGARG)
            HError(1219,"HLEd: Dictionary filename expected");
         dictFn = GetStrArg();
         break;
      case 'g':
         if (NextArg() != FLOATARG)
            HError(1219,"HLEd: Min fill gap expected (100ns units)");
         minGap = GetChkedFlt(0.0,100000000.0,s); 
         break;
      case 'i':
         if (NextArg()!=STRINGARG)
            HError(1219,"HLEd: Output MLF file name expected");
         if(SaveToMasterfile(GetStrArg())<SUCCESS)
            HError(1214,"HCopy: Cannot write to MLF");
         break;
      case 'l':
         if (NextArg()!=STRINGARG)
            HError(1219,"HLEd: Output label file directory expected");
         newDir = GetStrArg(); break;
      case 'm':
         LTriStrip(TRUE); break;
      case 'n':
         if (NextArg() != STRINGARG)
            HError(1219,"HLEd: New labels file name expected");
         newFn = GetStrArg(); break;
      case 'G':
         if (NextArg() != STRINGARG)
            HError(1219,"HLEd: Input label file format expected");
         if((ifmt = Str2Format(GetStrArg())) == ALIEN)
            HError(-1289,"HLEd: Warning ALIEN input label file format set");
         break;
      case 'I':
         if (NextArg() != STRINGARG)
            HError(1219,"HLEd: Input MLF file name expected");
         LoadMasterFile(GetStrArg()); break;
      case 'P':
         if (NextArg() != STRINGARG)
            HError(1219,"HLEd: Output label file format expected");
         if((ofmt = Str2Format(GetStrArg())) == ALIEN)
            HError(-1289,"HLEd: Warning ALIEN Label output file format set");
         break;
      case 'Q':
         Summary(); break;
      case 'T':
         trace = GetChkedInt(0,0777,s); break;
      case 'X':
         if (NextArg()!=STRINGARG)
            HError(1219,"HLEd: Output label file extension expected");
         newExt = GetStrArg(); break;
      default:
         HError(1219,"HLEd: Unknown switch %s",s);
      }
   }
   if (NextArg()!=STRINGARG)
      HError(1219,"HLEd: Edit script file name expected");
   scriptFn = GetStrArg();
   ReadScript(scriptFn);
   if (newFn!=NULL)
      if ((newLabs = fopen(newFn,"w")) == NULL)
         HError(1211,"HLEd: Cannot create new label file %s",newFn);
   if (dictFn!=NULL) {
      InitVocab(&vocab);
      if(ReadDict(dictFn,&vocab)<SUCCESS)
         HError(1213,"HLEd: ReadDict failed");
   }

   while (NumArgs()>0){
      if (NextArg()!=STRINGARG)
         HError(1219,"HLEd: 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)
               EditFile(me->pattern);
            me = me->next;
         }
      } else
         EditFile(labFn);
   }
   ListFinds();
   if (newLabs != NULL) fclose(newLabs);
   Exit(0);
   return (0);          /* never reached -- make compiler happy */
}

/* --------------------- Initialisation ----------------------- */

/* SetConfParms: set conf parms relevant to HLEd */
void SetConfParms(void)
{
   int i;
   
   nParm = GetConfig("HLED", TRUE, cParm, MAXGLOBS);
   if (nParm>0) {
      if (GetConfInt(cParm,nParm,"TRACE",&i)) trace = i;
   }
}

/* Initialise: confparms, str->int map and memory */
void Initialise(void)
{
   int i;
   char buf[MAXSTRLEN];
   LabId labid;
   
   SetConfParms();
   asterix = GetLabId("*",TRUE);
   for (i=1;i<=99;i++) {
      sprintf(buf,"%d",i);
      labid=GetLabId(buf,TRUE);
      labid->aux=(void*) i;
   }
   CreateHeap(&permHeap, "permHeap", MSTAK, 1, 1.2, 512, 4096);
   CreateHeap(&tempHeap, "tempHeap", MSTAK, 1, 1.2, 4096, 8192);
}

/* ------------------- Context List Handling ------------------ */

typedef struct _Context{
   short nDefs;
   LabId cname;
   LabId cdef[MAXARGS];
   struct _Context *next;
} Context;

static Context *xlist = NULL;

/* DefineContext:  add given context to list */
void DefineContext(LabId name, LabId *def, int nDefs)
{
   Context *p;

   p = (Context *)New(&permHeap,sizeof(Context));
   p->next = xlist; xlist = p;
   p->cname = name; p->nDefs = nDefs;
   memcpy(p->cdef,def,nDefs*sizeof(LabId));
   p->cdef[nDefs] = NULL;
}

/* PrintContexts: print all contexts */
void PrintContexts(void)
{
   Context *p = xlist;
   void PrintIdList(LabId *i);

   if (p == NULL){
      printf("No Contexts Defined\n");
      return;
   }
   printf("Contexts Defined:\n");
   while (p != NULL) {
      printf("   Context %s == ",p->cname->name);
      PrintIdList(p->cdef);
      printf("\n");
      p = p->next;
   }
}
   
/* GetContext: return list of ids defined by name */
LabId *GetContext(LabId name)
{
   Context *p = xlist;

   while (p != NULL) {
      if (p->cname == name)
         return p->cdef;
      p = p->next;
   }
   HError(1230,"GetContext: Context %s not defined by DC command",name->name);
   return NULL;
}

/* IsWBnd: return true if given id is an Interword boundary */
Boolean IsWBnd(LabId id)
{
   int i;
   
   for (i=0; i<nWB; i++)
      if (wbnd[i] == id) return TRUE;
   return FALSE;
}

/* PrintWBounds: print a list of defined interword boundaries */
void PrintWBounds(void)
{
   int i;
   
   if (nWB==0)
      printf("No inter-word boundaries defined\n");
   else {
      printf("Word Bounds Defined:\n");
      for (i=0;i<nWB;i++) printf("  %s",wbnd[i]->name);
      printf("\n");
   }
}

/* IsVBnd: return true if given id is a non-interword boundary */
Boolean IsVBnd(LabId id)
{
   int i;
   
   for (i=0; i<nVB; i++)
      if (vbnd[i] == id) return TRUE;
   return FALSE;
}

/* PrintVBounds: print a list of defined non-interword boundaries */
void PrintVBounds(void)
{
   int i;
   
   if (nVB==0)
      printf("No non inter-word boundaries defined\n");
   else {
      printf("Non Inter-Word Boundaries Defined:\n");
      for (i=0;i<nVB;i++) printf("  %s",vbnd[i]->name);
      printf("\n");
   }
}

/* ------------------- New Label Recording -------------------- */

/* PutLab: if given label new then output it to newLabs */
void PutLab(LabId id)
{
   if (id->aux == NULL) {  /* not seen this label before */
      id->aux = (void*) -1;
      fprintf(newLabs,"%s\n",id->name);
   }
}

/* ------------------- Find Accumulators ---------------------- */

typedef struct _FindAcc{
   LabId *pattern;      /* pattern to find */
   int acc;             /* num of matches */
   struct _FindAcc *next;
} FindAcc;

static FindAcc *facc = NULL;

/* GetFindAccumulator: return a pointer to acc for given pattern */
int *GetFindAccumulator(LabId *pat)
{
   FindAcc *p = facc;

   while (p != NULL) {
      if (p->pattern == pat)
         return &(p->acc);
      p = p->next;
   }
   HError(1292,"GetFindAccumulator: Acc %s %s %s not defined",
          pat[0]->name,pat[1]->name,pat[2]->name);
   return NULL;
}

/* MakeFindAccumulator: create a new find acc for given pattern */
void MakeFindAccumulator(LabId *pat)
{
   FindAcc *p;

   p = (FindAcc *)New(&permHeap,sizeof(FindAcc));
   p->next = facc; facc = p;
   p->pattern = pat; p->acc = 0;
}

/* ListFinds: list accumulated totals in all find accs */
void ListFinds(void)
{
   FindAcc *p = facc;
   LabId *i;

   if (p != NULL) {
      printf("Find Totals:\n");
      while (p != NULL){
         i = p->pattern;
         printf(" %4d  %s-%s+%s\n",p->acc,
                i[0]->name,i[1]->name,i[2]->name);
         p = p->next;
      }
   }
}

/* ---------------------- Print Script ---------------------- */

/* PrintIdList: print list of ids */
void PrintIdList(LabId *i)
{
   while (*i != NULL) {
      printf(" %s",(*i)->name);
      ++i;
   }
}

/* PrintScript: prints the currently loaded script - for tracing only */
void PrintScript(char *scriptFN)
{
   ScriptItem *i;
   int j;
   LabId src;

   printf("HLEd: script %s\n",scriptFN);