libsent/src/hmminfo/rdhmmdef_tiedmix.c

説明を見る。

/*
 * Copyright (c) 1991-2011 Kawahara Lab., Kyoto University
 * Copyright (c) 2000-2005 Shikano Lab., Nara Institute of Science and Technology
 * Copyright (c) 2005-2011 Julius project team, Nagoya Institute of Technology
 * All rights reserved
 */

#include <sent/stddefs.h>
#include <sent/htk_hmm.h>

extern char *rdhmmdef_token;    

static GCODEBOOK *
codebook_lookup(HTK_HMM_INFO *hmm, char *keyname)
{
  GCODEBOOK *book;

  if (hmm->codebook_root == NULL) return(NULL);
  book = aptree_search_data(keyname, hmm->codebook_root);
  if (book != NULL && strmatch(book->name, keyname)) {
    return book;
  } else {
    return NULL;
  }
}

void
codebook_add(HTK_HMM_INFO *hmm, GCODEBOOK *new)
{
  GCODEBOOK *match;
  if (hmm->codebook_root == NULL) {
    hmm->codebook_root = aptree_make_root_node(new, &(hmm->mroot));
  } else {
    match = aptree_search_data(new->name, hmm->codebook_root);
    if (match != NULL && strmatch(match->name, new->name)) {
      jlog("Error: rdhmmdef_tiedmix: ~s \"%s\" is already defined\n", new->name);
      rderr(NULL);
    } else {
      aptree_add_entry(new->name, new, match->name, &(hmm->codebook_root), &(hmm->mroot));
    }
  }
}

static void
tmix_create_codebook_index(HTK_HMM_INFO *hmminfo, GCODEBOOK *book)
{
  char *mixname;
  HTK_HMM_Dens *dtmp;
  int i;
  int realbooknum = 0;

  mixname = (char *)mymalloc(strlen(book->name)+30);
  book->d = (HTK_HMM_Dens **) mybmalloc2(sizeof(HTK_HMM_Dens *) * book->num, &(hmminfo->mroot));
  for (i=0;i<book->num;i++) {
    sprintf(mixname, "%s%d", book->name, i + 1);
    if ((dtmp = dens_lookup(hmminfo, mixname)) == NULL) {
/* 
 *       jlog("Error: mixture \"%s\" (%dth mixture in codebook \"%s\") not found\n", mixname, i + 1, book->name);
 *       rderr(NULL);
 */
      book->d[i] = NULL;
    } else {
      book->d[i] = dtmp;
      realbooknum++;
    }
  }
  if (realbooknum < book->num) {
    jlog("Warning: rdhmmdef_tiedmix: book [%s]: defined=%d < %d\n",
         book->name, realbooknum, book->num);
  }
  
  free(mixname);
}

void
tmix_read(FILE *fp, HTK_HMM_PDF *mpdf, HTK_HMM_INFO *hmm)
{
  char *bookname;
  GCODEBOOK *thebook;
  int mid, i;

  NoTokErr("missing TMIX bookname");
  bookname = rdhmmdef_token;
  /* check whether the specified codebook exist */
  if ((thebook = codebook_lookup(hmm, bookname)) == NULL) {
    /* create GCODEBOOK global index structure from mixture macros */
    thebook = (GCODEBOOK *)mybmalloc2(sizeof(GCODEBOOK), &(hmm->mroot));
    thebook->name = mybstrdup2(bookname, &(hmm->mroot));
    thebook->num = mpdf->mix_num;
    /* map codebook id to HTK_HMM_Dens* */
    tmix_create_codebook_index(hmm, thebook);
    /* register the new codebook */
    codebook_add(hmm, thebook);
    thebook->id = hmm->codebooknum;
    hmm->codebooknum++;
    /* set maximum codebook size */
    if (hmm->maxcodebooksize < thebook->num) hmm->maxcodebooksize = thebook->num;
  } else {
    /* check coherence */
    if (mpdf->mix_num != thebook->num) {
      rderr("tmix_read: TMIX weight num don't match the codebook size");
    }
  }

  /* set pointer to the GCODEBOOK structure  */
  mpdf->b = (HTK_HMM_Dens **)thebook;

  /* store the weights to `mpdf->bweight[]' */
  read_token(fp);
  mpdf->bweight = (PROB *) mybmalloc2(sizeof(PROB) * mpdf->mix_num, &(hmm->mroot));
  {
    int len;
    double w;

    mid = 0;
    while (mid < mpdf->mix_num)
    {
      char *p, q;
      NoTokErr("missing some TMIX weights");
      if ((p = strchr(rdhmmdef_token, '*')) == NULL) {
        len = 1;
        w = atof(rdhmmdef_token);
      } else {
        len = atoi(p+1);
        q = *p;
        *p = '\0';
        w = atof(rdhmmdef_token);
        *p = q;
      }
      read_token(fp);
      for(i=0;i<len;i++) {
        mpdf->bweight[mid] = (PROB)log(w);
        mid++;
      }
    }
  }

  /* mark info as tied mixture */
  hmm->is_tied_mixture = TRUE;
}