spsegment.c

julius version 4.12.about sound recognition.

 * input will be shrinked to the segmented length and the unprocessed
 * region are copied to rest_param for the next decoding.
 * 
 * @param recog [in] engine instance
 * </EN>
 * @callgraph
 * @callergraph
 */
void
finalize_segment(Recog *recog)
{
  int spstart;
  RecogProcess *r;
  MFCCCalc *mfcc;
  boolean ok_p;

  /* トレリス幌姜眉における呵锑帽胳を妈2パスの幌姜眉帽胳として呈羌 */
  /* fix initial/last word hypothesis of the next 2nd pass to the best
     word hypothesis at the first/last frame in backtrellis*/
  for(r=recog->process_list;r;r=r->next) {
    if (!r->live) continue;
    if (r->lmtype == LM_PROB) {
      set_terminal_words(r);
    }
  }

  /* パラメ〖タを, 海妈１パスが姜位したセグメント惰粗と荒りの惰粗に尸充する. 
     ただし儡鲁婶のsp惰粗婶尸(sparea_start..len-1)は≈のりしろ∽として尉数に
     コピ〖する */
  /* Divide input parameter into two: the last segment and the rest.
     The short-pause area (sparea_start..len-1) is considered as "tab",
     copied in both parameters
   */
  /* param[sparea_start..framelen] -> rest_param
     param[0..len-1] -> param
     [sparea_start...len-1] overlapped
  */

  ok_p = FALSE;
  for (mfcc = recog->mfcclist; mfcc; mfcc = mfcc->next) {
    if (mfcc->segmented) {
      spstart = mfcc->sparea_start;
      ok_p = TRUE;
      break;
    }
  }

  if (ok_p) {
    /* the input was segmented in an instance */
    /* shrink all param the len and store restart parameters in rest_param */
    /* for each mfcc */
    if (verbose_flag) jlog("STAT: segmented: next decoding will restart from %d\n", spstart);

    for (mfcc = recog->mfcclist; mfcc; mfcc = mfcc->next) {
      if (verbose_flag) jlog("STAT: MFCC%02d: segmented: processed length=%d\n", mfcc->id, mfcc->last_time);

      /* copy the rest to mfcc->rest_param and shrink mfcc->param */
      mfcc_copy_to_rest_and_shrink(mfcc, spstart, mfcc->last_time);
    }

    /* reset last_word info */
    for(r=recog->process_list;r;r=r->next) {
      if (!r->live) continue;
      r->sp_break_last_nword_allow_override = TRUE;
    }
    
  } else {
    
    /* last segment is on end of input: no rest parameter */
    for (mfcc = recog->mfcclist; mfcc; mfcc = mfcc->next) {
      mfcc->rest_param = NULL;
    }
    /* reset last_word info */
    for(r=recog->process_list;r;r=r->next) {
      if (!r->live) continue;
      r->sp_break_2_begin_word = WORD_INVALID;
      r->sp_break_last_word = WORD_INVALID;
      r->sp_break_last_nword = WORD_INVALID;
      r->sp_break_last_nword_allow_override = FALSE;
    }
  }
}

#ifdef BACKEND_VAD

/** 
 * <EN>
 * Initialize parameters for decoder/GMM-based VAD.
 * This will be called before recognition start for each segment.
 * </EN>
 * <JA>
 * Decode/GMM-based VAD のためにパラメ〖タを介袋步する. 
 * 称掐蜗セグメントの千急借妄を幌める涟に钙ばれる. 
 * </JA>
 * 
 * @param recog [i/o] engine instance
 * 
 * @callgraph
 * @callergraph
 */
void
spsegment_init(Recog *recog)
{
  RecogProcess *p;
  /* at first time, recognition does not start yet */
#ifdef SPSEGMENT_NAIST
  for(p=recog->process_list;p;p=p->next) {
    p->pass1.after_trigger = FALSE;
    p->pass1.trigger_duration = 0;
  }
#endif
#ifdef GMM_VAD
  if (recog->gmm) {
    recog->gc->after_trigger = FALSE;
    recog->gc->duration = 0;
  }
#endif
  recog->triggered = FALSE;
}

/** 
 * <EN>
 * @brief  Detect speech up-trigger and synhronize among instances.
 * 
 * This function inspects all recognition instancces and gmm components
 * to see if any of them has detected trigger up (beginning of speech)
 * at the last recognition process.  If trigger has been detected,
 * set trigger-up status for all the instances.
 * </EN>
 * <JA>
 * @brief  不兰惰粗倡幌の浮叫およびインスタンス粗票袋. 
 * 
 * 链ての千急借妄インスタンスとGMM借妄婶について·木涟の千急借妄で
 * トリガアップ∈不兰惰粗倡幌∷が冉年されたかどうかを拇べる. 
 * 倡幌された眷圭は·链ての千急借妄インスタンスでアップトリガをマ〖クする. 
 * </JA>
 * 
 * @param recog [in] engine instance
 * 
 * @return TRUE if triggered, or FALSE if not.
 * 
 * @callgraph
 * @callergraph
 */
boolean
spsegment_trigger_sync(Recog *recog)
{
  RecogProcess *p;
  boolean ok_p;

  ok_p = FALSE;
  if (recog->jconf->decodeopt.segment) {
#ifdef SPSEGMENT_NAIST
    for(p = recog->process_list; p; p = p->next) {
      if (!p->live) continue;
      if (p->pass1.after_trigger) {
	ok_p = TRUE;
	break;
      }
    }
#endif
#ifdef GMM_VAD
    if (recog->gmm) {
      if (recog->gc->after_trigger) {
	ok_p = TRUE;
      }
    }
#endif
  }
  if (ok_p) {
    /* up trigger detected */
#ifdef SPSEGMENT_NAIST
    for(p = recog->process_list; p; p = p->next) {
      if (!p->live) continue;
      p->pass1.after_trigger = TRUE;
    }
#endif
#ifdef GMM_VAD
    if (recog->gmm) {
      recog->gc->after_trigger = TRUE;
    }
#endif
  }
  
  return ok_p;
}

#endif /* BACKEND_VAD */

/** 
 * <EN>
 * @brief  Check if rewind and restart of recognition is needed.
 *
 * This function checks if an instance requires rewinding of input
 * samples, and if recognition re-processing is needed after rewinding.
 * 
 * </EN>
 * <JA>
 * @brief  船き提しと千急浩倡の涩妥拉をチェックする. 
 * 
 * 不兰千急借妄において船き提しが涩妥がどうか拇べ·涩妥な眷圭は
 * フレ〖ム眶と·船き提した稿に船提し尸の千急借妄を乖うかどうかを手す. 
 * </JA>
 * 
 * @param recog [in] engine instance
 * @param rf_ret [out] length of frame to rewind
 * @param repro_ret [out] TRUE if re-process is required after rewinding
 * 
 * @return TRUE if rewinding is required, or FALSE if not.
 * 
 * @callgraph
 * @callergraph
 */
boolean
spsegment_need_restart(Recog *recog, int *rf_ret, boolean *repro_ret)
{
#ifdef SPSEGMENT_NAIST
  RecogProcess *p;
#endif
  boolean ok_p;
  int rewind_frame;
  boolean reprocess;

  ok_p = FALSE;
  if (recog->jconf->decodeopt.segment) {
#ifdef SPSEGMENT_NAIST
    /* check for rewind request from each process */
    for(p = recog->process_list; p; p = p->next) {
      if (!p->live) continue;
      if (p->pass1.want_rewind) {
	p->pass1.want_rewind = FALSE;
	rewind_frame = p->pass1.rewind_frame;
	reprocess = p->pass1.want_rewind_reprocess;
	ok_p = TRUE;
	break;
      }
    }
#endif /* SPSEGMENT_NAIST */
#ifdef GMM_VAD
    if (recog->gmm) {
      if (recog->gc->want_rewind) {
	recog->gc->want_rewind = FALSE;
#ifdef SPSEGMENT_NAIST
	/* set to earlier one */
	if (rewind_frame > recog->gc->rewind_frame) rewind_frame = recog->gc->rewind_frame;
#else
	rewind_frame = recog->gc->rewind_frame;
#endif
	reprocess = recog->gc->want_rewind_reprocess;
	ok_p = TRUE;
      }
    }
#endif
    *rf_ret = rewind_frame;
    *repro_ret = reprocess;
  }

  return(ok_p);
}

/** 
 * <EN>
 * @brief  Execute rewinding.
 *
 * This function will set re-start point for the following processing,
 * and shrink the parameters for the rewinded part.  The re-start point
 * is 0 (beginning of rest samples) for recognition restart, or
 * simply go back to the specified rewind frames for non restart.
 * 
 * </EN>
 * <JA>
 * @brief  船き提し借妄
 *
 * 肌搀の掐蜗借妄の倡幌爬を疯年し·船き提し尸パラメ〖タを低める. 
 * 浩倡回年の眷圭倡幌爬はパラメ〖タの黎片に·それ笆嘲の眷圭は船提した
 * 尸だけ提った疤弥にセットされる. 
 * 
 * </JA>
 * 
 * @param recog [i/o] engine instance
 * @param rewind_frame [in] frame length to rewind
 * @param reprocess [in] TRUE if re-processing recognition is required for the following processing
 * 
 * @callgraph
 * @callergraph
 */
void
spsegment_restart_mfccs(Recog *recog, int rewind_frame, boolean reprocess)
{
  MFCCCalc *mfcc;

  for (mfcc = recog->mfcclist; mfcc; mfcc = mfcc->next) {
    if (!mfcc->valid) continue;
    /* set last segmented time */
    mfcc->last_time = mfcc->f - 1;
    /* reset frame pointers */
    if (reprocess) {
      /* set all mfcc to initial point for re-process the whole frames */
      mfcc->f = -1;
    } else {
      /* just bring back to the new last point after shrink */
      mfcc->f -= rewind_frame;
    }
    /* shrink the current mfcc */
    mfcc_shrink(mfcc, rewind_frame);
  }
}

/* end of file */