corpus.c

CMU大名鼎鼎的SPHINX－3大词汇量连续语音识别系统

	    k = sscanf (line, "%s %d %d %s", uttfile, sf, ef, uttid);
    } while (k <= 0);
    
    if ((k == 2) || ( (k >= 3) && ((*sf >= *ef) || (*sf < 0))) )
	E_FATAL("Error in ctlfile: %s\n", line);
    
    if (k < 4) {
	/* Create utt-id from mfc-filename (and sf/ef if specified) */
	path2basename (uttfile, base);
	strip_fileext (base, uttid);
	
	if (k == 3) {
	    k = strlen(uttid);
	    sprintf (uttid+k, "_%d_%d", *sf, *ef);
	} else {
	    *sf = 0;
	    *ef = -1;	/* Signifies "until EOF" */
	}
    }
    
    return 0;
}


ptmr_t ctl_process (char *ctlfile, char* ctlmllrfile, int32 nskip, int32 count,
		    void (*func) (void *kb, char *uttfile, int32 sf, int32 ef, char *uttid),
		    void *kb)
{
  FILE *fp, *mllrfp;
  char uttfile[16384], uttid[4096];
  char regmatfile[4096];
  int32 sf, ef;
  ptmr_t tm;
  
  mllrfp = NULL;
  E_INFO("Batch mode recognition without dynamic LM\n");

  if (ctlfile) {
    if ((fp = fopen(ctlfile, "r")) == NULL)
      E_FATAL_SYSTEM("fopen(%s,r) failed\n", ctlfile);
  } else
    fp = stdin;

  if(ctlmllrfile){
    if ((mllrfp = fopen(ctlmllrfile, "r")) == NULL)
      E_FATAL_SYSTEM("fopen(%s,r) failed\n", ctlmllrfile);
  }

  ptmr_init (&tm);
  
  if (nskip > 0) {
    E_INFO("Skipping %d entries at the beginning of %s\n", nskip, ctlfile);
    
    for (; nskip > 0; --nskip) {
      if (ctl_read_entry (fp, uttfile, &sf, &ef, uttid) < 0) {
	fclose (fp);
	return tm;
      }
    }

    if(ctlmllrfile){
      for (; nskip > 0; --nskip) {
	if (ctl_read_entry (fp, regmatfile, &sf, &ef, uttid) < 0) {
	  E_ERROR("MLLR cannot be read when skipping the %d-th sentence\n",nskip);
	  fclose (fp);
	  return tm;
	}
      }
    }
  }
  
  for (; count > 0; --count) {
    if (ctl_read_entry (fp, uttfile, &sf, &ef, uttid) < 0)
      break;

    if(ctlmllrfile){
      if (ctl_read_entry (mllrfp, regmatfile, &sf, &ef, uttid) < 0){
	E_ERROR("MLLR cannot be read when counting the %d-th sentence\n",count);
	break;
      }
    }
    
    /* Process this utterance */
    ptmr_start (&tm);
    if (func){
      if(ctlmllrfile) kb_setmllr(regmatfile,kb);
      (*func)(kb, uttfile, sf, ef, uttid);
    }
    ptmr_stop (&tm);
    
    E_INFO("%s: %6.1f sec CPU, %6.1f sec Clk;  TOT: %8.1f sec CPU, %8.1f sec Clk\n\n",
	   uttid, tm.t_cpu, tm.t_elapsed, tm.t_tot_cpu, tm.t_tot_elapsed);


    ptmr_reset (&tm);
  }
  
  fclose (fp);
  
  return tm;
}

ptmr_t ctl_process_dyn_lm (char *ctlfile, char *ctllmfile, char* ctlmllrfile, int32 nskip, int32 count,
		    void (*func) (void *kb, char *uttfile, int32 sf, int32 ef, char *uttid),
		    void *kb)
{
  FILE *fp;
  FILE *ctllmfp;
  FILE *ctlmllrfp;
  char uttfile[16384], uttid[4096];
  char lmname[4096];
  char regmatname[4096];
  char tmp[4096];
  int32 sf, ef;
  ptmr_t tm;
  
  ctllmfp=NULL;
  ctlmllrfp=NULL;
  E_INFO("Batch mode recognition with dynamic LM\n");
  if (ctlfile) {
    if ((fp = fopen(ctlfile, "r")) == NULL)
      E_FATAL_SYSTEM("fopen(%s,r) failed\n", ctlfile);
  } else
    fp = stdin;

  if(ctllmfile){
    if ((ctllmfp = fopen(ctllmfile, "r")) == NULL)
      E_FATAL_SYSTEM("fopen(%s,r) failed\n", ctllmfile);
  }

  if(ctlmllrfile) {
    E_INFO("MLLR is used in this session\n");
    if((ctlmllrfp = fopen(ctlmllrfile, "r")) == NULL)
      E_FATAL_SYSTEM("fopen(%s,r) failed\n", ctlmllrfile);
  }

  ptmr_init (&tm);
  
  if (nskip > 0) {
    E_INFO("Skipping %d entries at the beginning of %s\n", nskip, ctlfile);
    
    for (; nskip > 0; --nskip) {
      if (ctl_read_entry (fp, uttfile, &sf, &ef, uttid) < 0) {
	fclose (fp);
	return tm;
      }

      /*This checks the size of the control file of the lm in batch mode*/
      if (ctl_read_entry (ctllmfp, lmname, &sf, &ef, tmp) < 0) {
	fclose (ctllmfp);
	E_ERROR("LM cannot be read when skipping the %d-th sentence\n",nskip);
	return tm;
      }

      /*This checks the size of the control file of the mllr in batch mode */
      if(ctlmllrfile){
	if (ctl_read_entry (ctlmllrfp, regmatname, &sf, &ef, tmp) < 0) {
	  fclose (ctlmllrfp);
	  E_ERROR("MLLR cannot be read when skipping the %d-th sentence\n",nskip);
	  return tm;
	}
      }

    }
  }
  
  for (; count > 0; --count) {
    if (ctl_read_entry (fp, uttfile, &sf, &ef, uttid) < 0)
      break;

    /*This checks the size of the control file in batch mode*/
    if (ctl_read_entry (ctllmfp, lmname, &sf, &ef, tmp) < 0) {
      fclose (ctllmfp);
      E_ERROR("LM cannot be read when counting the %d-th sentence\n",count);
      return tm;
    }

    if(ctlmllrfile){
      if (ctl_read_entry (ctlmllrfp, regmatname, &sf, &ef, tmp) < 0) {
	fclose (ctlmllrfp);
	E_ERROR("MLLR cannot be read when counting the %d-th sentence\n",count);
	return tm;
      }
    }
    

    /* Process this utterance */
    ptmr_start (&tm);
    E_INFO("filename %s, lmname %s\n",uttfile,lmname);
    if (func){
      kb_setlm(lmname,kb);
      if(ctlmllrfile) kb_setmllr(regmatname,kb);
      (*func)(kb, uttfile, sf, ef, uttid);
    }
    ptmr_stop (&tm);
    
    E_INFO("%s: %6.1f sec CPU, %6.1f sec Clk;  TOT: %8.1f sec CPU, %8.1f sec Clk\n\n",
	   uttid, tm.t_cpu, tm.t_elapsed, tm.t_tot_cpu, tm.t_tot_elapsed);


    ptmr_reset (&tm);
  }
  
  fclose (fp);
  fclose (ctllmfp);
  fclose (ctlmllrfp);
  
  return tm;
}



ptmr_t ctl_process_utt (char *uttfile, int32 count,
			void (*func) (void *kb, char *uttfile, int32 sf, int32 ef, char *uttid),
			void *kb)
{
    char uttid[4096];
    char base[16384];
    int32 i, c;
    int32 ts, newts;
    ptmr_t tm;
    
    ptmr_init (&tm);
    path2basename (uttfile, base);
    strip_fileext (base, uttid);
    strcpy (base, uttid);
    
    ts = -1;
    for (c = 0; c < count; c++) {
	/* Wait for uttfile to change from previous iteration */
	for (i = 0;; i++) {
	    newts = stat_mtime (uttfile);
	    if ((newts >= 0) && (newts != ts))
		break;
	    
	    if (i == 0)
		E_INFO("Waiting for %s, count %d, c %d\n", uttfile,count,c);
	    SLEEP_SEC (1);
	}
	ts = newts;
	
	/* Form uttid */
	sprintf (uttid, "%s_%08d", base, c);
	
	/* Process this utterance */
	ptmr_start (&tm);
	if (func)
	    (*func)(kb, uttfile, 0, -1, uttid);
	ptmr_stop (&tm);
	
	E_INFO("%s: %6.1f sec CPU, %6.1f sec Clk;  TOT: %8.1f sec CPU, %8.1f sec Clk\n\n",
	       uttid, tm.t_cpu, tm.t_elapsed, tm.t_tot_cpu, tm.t_tot_elapsed);
	
	ptmr_reset (&tm);
    }
    
    return tm;
}


void ctl_infile (char *file, char *dir, char *ext, char *utt)
{
    int32 l1, l2;
    
    assert (utt);
    
    if (ext && (ext[0] != '\0')) {
	l1 = strlen(ext);
	l2 = strlen(utt);
	if ((l2 > l1) && (utt[l2-l1-1] == '.') && (strcmp (utt+(l2-l1), ext) == 0))
	    ext = NULL;		/* utt already has the desired extension */
    }
    
    if ((utt[0] != '/') && dir) {
	/* Dir specified for relative uttfile pathname */
	if (ext && (ext[0] != '\0'))
	    sprintf (file, "%s/%s.%s", dir, utt, ext);
	else
	    sprintf (file, "%s/%s", dir, utt);
    } else {
	if (ext && (ext[0] != '\0'))
	    sprintf (file, "%s.%s", utt, ext);
	else
	    strcpy (file, utt);
    }
}


void ctl_outfile (char *file, char *dir, char *ext, char *utt, char *uttid)
{
    int32 k;
    
    k = strlen(dir);
    
    if ((k > 4) && (strcmp (dir+k-4, ",CTL") == 0)) {	/* HACK!! Hardwired ,CTL */
	if (utt[0] != '/') {
	    strcpy (file, dir);
	    file[k-4] = '/';
	    strcpy (file+k-3, utt);
	} else
	    strcpy (file, utt);
    } else {
	strcpy (file, dir);
	file[k] = '/';
	strcpy (file+k+1, uttid);
    }
    
    if (ext && (ext[0] != '\0')) {
	strcat (file, ".");
	strcat (file, ext);
    }
}