void
mdef_sseq2sen_active(mdef_t * mdef, uint8 * sseq, uint8 * sen)
{
int32 ss, i;
s3senid_t *sp;
for (ss = 0; ss < mdef_n_sseq(mdef); ss++) {
if (sseq[ss]) {
sp = mdef->sseq[ss];
for (i = 0; i < mdef_n_emit_state(mdef); i++)
sen[sp[i]] = 1;
}
}
}
void dict2pid_comsseq2sen_active (dict2pid_t *d2p, mdef_t *mdef, int32 *comssid, int32 *sen)
{
int32 ss, cs, i, j;
s3senid_t *csp, *sp; /* Composite state pointer */
for (ss = 0; ss < d2p->n_comsseq; ss++) {
if (comssid[ss]) {
csp = d2p->comsseq[ss];
for (i = 0; i < mdef_n_emit_state(mdef); i++) {
cs = csp[i];
sp = d2p->comstate[cs];
for (j = 0; IS_S3SENID(sp[j]); j++)
sen[sp[j]] = 1;
}
}
}
}
int
srch_FLAT_FWD_init(kb_t * kb, /**< The KB */
void *srch /**< The pointer to a search structure */
)
{
srch_FLAT_FWD_graph_t *fwg;
kbcore_t *kbc;
srch_t *s;
mdef_t *mdef;
dict_t *dict;
lm_t *lm;
kbc = kb->kbcore;
s = (srch_t *) srch;
mdef = kbcore_mdef(kbc);
dict = kbcore_dict(kbc);
lm = kbcore_lm(kbc);
fwg = ckd_calloc(1, sizeof(srch_FLAT_FWD_graph_t));
E_INFO("Initialization\n");
/** For convenience */
fwg->kbcore = s->kbc;
/* Allocate whmm structure */
fwg->hmmctx = hmm_context_init(mdef_n_emit_state(mdef),
kbcore_tmat(kbc)->tp, NULL,
mdef->sseq);
fwg->whmm = (whmm_t **) ckd_calloc(dict->n_word, sizeof(whmm_t *));
/* Data structures needed during word transition */
/* These five things need to be tied into the same structure. Such that when multiple LM they could be switched. */
fwg->rcscore = NULL;
fwg->rcscore = (int32 *) ckd_calloc(mdef->n_ciphone, sizeof(int32));
fwg->ug_backoff =
(backoff_t *) ckd_calloc(mdef->n_ciphone, sizeof(backoff_t));
fwg->filler_backoff =
(backoff_t *) ckd_calloc(mdef->n_ciphone, sizeof(backoff_t));
fwg->tg_trans_done = (uint8 *) ckd_calloc(dict->n_word, sizeof(uint8));
fwg->word_ugprob = init_word_ugprob(mdef, lm, dict);
/* Input candidate-word lattices information to restrict search; if any */
fwg->word_cand_dir = cmd_ln_str_r(kbcore_config(fwg->kbcore), "-inlatdir");
fwg->latfile_ext = cmd_ln_str_r(kbcore_config(fwg->kbcore), "-latext");
fwg->word_cand_win = cmd_ln_int32_r(kbcore_config(fwg->kbcore), "-inlatwin");
if (fwg->word_cand_win < 0) {
E_ERROR("Invalid -inlatwin argument: %d; set to 50\n",
fwg->word_cand_win);
fwg->word_cand_win = 50;
}
/* Allocate pointers to lists of word candidates in each frame */
if (fwg->word_cand_dir) {
fwg->word_cand =
(word_cand_t **) ckd_calloc(S3_MAX_FRAMES,
sizeof(word_cand_t *));
fwg->word_cand_cf =
(s3wid_t *) ckd_calloc(dict->n_word + 1, sizeof(s3wid_t));
}
/* Initializing debugging information such as trace_wid,
word_dump_sf, word_dump_ef, hmm_dump_sf and hmm_dump_ef */
fwg->fwdDBG = init_fwd_dbg(fwg);
fwg->ctr_mpx_whmm = pctr_new("mpx");
fwg->ctr_nonmpx_whmm = pctr_new("~mpx");
fwg->ctr_latentry = pctr_new("lat");
/** Initialize the context table */
fwg->ctxt = ctxt_table_init(kbcore_dict(kbc), kbcore_mdef(kbc));
/* Viterbi history structure */
fwg->lathist = latticehist_init(cmd_ln_int32_r(kbcore_config(fwg->kbcore), "-bptblsize"),
S3_MAX_FRAMES + 1);
/* Glue the graph structure */
s->grh->graph_struct = fwg;
s->grh->graph_type = GRAPH_STRUCT_FLAT;
return SRCH_SUCCESS;
}
void dict2pid_dump (FILE *fp, dict2pid_t *d2p, mdef_t *mdef, dict_t *dict)
{
int32 w, p, pronlen;
int32 i, j, b, l, r;
fprintf (fp, "# INTERNAL (wd comssid ssid ssid ... ssid comssid)\n");
for (w = 0; w < dict_size(dict); w++) {
fprintf (fp, "%30s ", dict_wordstr(dict, w));
pronlen = dict_pronlen(dict, w);
for (p = 0; p < pronlen; p++)
fprintf (fp, " %5d", d2p->internal[w][p]);
fprintf (fp, "\n");
}
fprintf (fp, "#\n");
fprintf (fp, "# LDIPH_LC (b r l ssid)\n");
for (b = 0; b < mdef_n_ciphone(mdef); b++) {
for (r = 0; r < mdef_n_ciphone(mdef); r++) {
for (l = 0; l < mdef_n_ciphone(mdef); l++) {
if (IS_S3SSID(d2p->ldiph_lc[b][r][l]))
fprintf (fp, "%6s %6s %6s %5d\n",
mdef_ciphone_str (mdef, (s3cipid_t)b),
mdef_ciphone_str (mdef, (s3cipid_t)r),
mdef_ciphone_str (mdef, (s3cipid_t)l),
d2p->ldiph_lc[b][r][l]); /* RAH, ldiph_lc is returning an int32, %d expects an int16 */
}
}
}
fprintf (fp, "#\n");
fprintf (fp, "# SINGLE_LC (b l comssid)\n");
for (b = 0; b < mdef_n_ciphone(mdef); b++) {
for (l = 0; l < mdef_n_ciphone(mdef); l++) {
if (IS_S3SSID(d2p->single_lc[b][l]))
fprintf (fp, "%6s %6s %5d\n",
mdef_ciphone_str (mdef, (s3cipid_t)b),
mdef_ciphone_str (mdef, (s3cipid_t)l),
d2p->single_lc[b][l]); /* RAH, single_lc is returning an int32, %d expects an int16 */
}
}
fprintf (fp, "#\n");
fprintf (fp, "# SSEQ %d (senid senid ...)\n", mdef->n_sseq);
for (i = 0; i < mdef->n_sseq; i++) {
fprintf (fp, "%5d ", i);
for (j = 0; j < mdef_n_emit_state(mdef); j++)
fprintf (fp, " %5d", mdef->sseq[i][j]);
fprintf (fp, "\n");
}
fprintf (fp, "#\n");
fprintf (fp, "# COMSSEQ %d (comstate comstate ...)\n", d2p->n_comsseq);
for (i = 0; i < d2p->n_comsseq; i++) {
fprintf (fp, "%5d ", i);
for (j = 0; j < mdef_n_emit_state(mdef); j++)
fprintf (fp, " %5d", d2p->comsseq[i][j]);
fprintf (fp, "\n");
}
fprintf (fp, "#\n");
fprintf (fp, "# COMSTATE %d (senid senid ...)\n", d2p->n_comstate);
for (i = 0; i < d2p->n_comstate; i++) {
fprintf (fp, "%5d ", i);
for (j = 0; IS_S3SENID(d2p->comstate[i][j]); j++)
fprintf (fp, " %5d", d2p->comstate[i][j]);
fprintf (fp, "\n");
}
fprintf (fp, "#\n");
fprintf (fp, "# END\n");
fflush (fp);
}
/*
* Convert the glist of ssids to a composite sseq id. Return the composite ID.
*/
static s3ssid_t ssidlist2comsseq (glist_t g, mdef_t *mdef, dict2pid_t *dict2pid,
hash_table_t *hs, /* For composite states */
hash_table_t *hp) /* For composite senone seq */
{
int32 i, j, n, s, ssid;
s3senid_t **sen;
s3senid_t *comsenid;
gnode_t *gn;
n = glist_count (g);
if (n <= 0)
E_FATAL("Panic: length(ssidlist)= %d\n", n);
/* Space for list of senones for each state, derived from the given glist */
sen = (s3senid_t **) ckd_calloc (mdef_n_emit_state (mdef), sizeof(s3senid_t *));
for (i = 0; i < mdef_n_emit_state (mdef); i++) {
sen[i] = (s3senid_t *) ckd_calloc (n+1, sizeof(s3senid_t));
sen[i][0] = BAD_S3SENID; /* Sentinel */
}
/* Space for composite senone ID for each state position */
comsenid = (s3senid_t *) ckd_calloc (mdef_n_emit_state (mdef), sizeof(s3senid_t));
for (gn = g; gn; gn = gnode_next(gn)) {
ssid = gnode_int32 (gn);
/* Expand ssid into individual states (senones); insert in sen[][] if not present */
for (i = 0; i < mdef_n_emit_state (mdef); i++) {
s = mdef->sseq[ssid][i];
for (j = 0; (IS_S3SENID(sen[i][j])) && (sen[i][j] != s); j++);
if (NOT_S3SENID(sen[i][j])) {
sen[i][j] = s;
sen[i][j+1] = BAD_S3SENID;
}
}
}
/* Convert senones list for each state position into composite state */
for (i = 0; i < mdef_n_emit_state (mdef); i++) {
for (j = 0; IS_S3SENID(sen[i][j]); j++);
assert (j > 0);
j = hash_enter_bkey (hs, (char *)(sen[i]), j*sizeof(s3senid_t), dict2pid->n_comstate);
if (j == dict2pid->n_comstate)
dict2pid->n_comstate++; /* New composite state */
else
ckd_free ((void *) sen[i]);
comsenid[i] = j;
}
ckd_free (sen);
/* Convert sequence of composite senids to composite sseq ID */
j = hash_enter_bkey (hp, (char *)comsenid, mdef->n_emit_state * sizeof(s3senid_t),
dict2pid->n_comsseq);
if (j == dict2pid->n_comsseq) {
dict2pid->n_comsseq++;
if (dict2pid->n_comsseq >= MAX_S3SENID)
E_FATAL("#Composite sseq limit(%d) reached; increase MAX_S3SENID\n",
dict2pid->n_comsseq);
} else
ckd_free ((void *) comsenid);
return ((s3ssid_t)j);
}