static int32 dict_read (FILE *fp, dict_t *d)
{
char line[16384], **wptr;
s3cipid_t p[4096];
int32 lineno, nwd;
s3wid_t w;
int32 i, maxwd;
maxwd = 4092;
wptr = (char **) ckd_calloc (maxwd, sizeof(char *));
lineno = 0;
while (fgets (line, sizeof(line), fp) != NULL) {
lineno++;
if (line[0] == '#') /* Comment line */
continue;
if ((nwd = str2words (line, wptr, maxwd)) < 0)
E_FATAL("str2words(%s) failed; Increase maxwd from %d\n", line, maxwd);
if (nwd == 0) /* Empty line */
continue;
/* wptr[0] is the word-string and wptr[1..nwd-1] the pronunciation sequence */
if (nwd == 1) {
E_ERROR("Line %d: No pronunciation for word %s; ignored\n", lineno, wptr[0]);
continue;
}
/* Convert pronunciation string to CI-phone-ids */
for (i = 1; i < nwd; i++) {
p[i-1] = dict_ciphone_id (d, wptr[i]);
if (NOT_CIPID(p[i-1])) {
E_ERROR("Line %d: Bad ciphone: %s; word %s ignored\n",
lineno, wptr[i], wptr[0]);
break;
}
}
if (i == nwd) { /* All CI-phones successfully converted to IDs */
w = dict_add_word (d, wptr[0], p, nwd-1);
if (NOT_WID(w))
E_ERROR("Line %d: dict_add_word (%s) failed (duplicate?); ignored\n",
lineno, wptr[0]);
}
}
ckd_free (wptr);
return 0;
}
static int32
dict_read(FILE * fp, dict_t * d)
{
lineiter_t *li;
char **wptr;
s3cipid_t *p;
int32 lineno, nwd;
s3wid_t w;
int32 i, maxwd;
size_t stralloc, phnalloc;
maxwd = 512;
p = (s3cipid_t *) ckd_calloc(maxwd + 4, sizeof(*p));
wptr = (char **) ckd_calloc(maxwd, sizeof(char *)); /* Freed below */
lineno = 0;
stralloc = phnalloc = 0;
for (li = lineiter_start(fp); li; li = lineiter_next(li)) {
lineno++;
if (0 == strncmp(li->buf, "##", 2)
|| 0 == strncmp(li->buf, ";;", 2))
continue;
if ((nwd = str2words(li->buf, wptr, maxwd)) < 0) {
/* Increase size of p, wptr. */
nwd = str2words(li->buf, NULL, 0);
assert(nwd > maxwd); /* why else would it fail? */
maxwd = nwd;
p = (s3cipid_t *) ckd_realloc(p, (maxwd + 4) * sizeof(*p));
wptr = (char **) ckd_realloc(wptr, maxwd * sizeof(*wptr));
}
if (nwd == 0) /* Empty line */
continue;
/* wptr[0] is the word-string and wptr[1..nwd-1] the pronunciation sequence */
if (nwd == 1) {
E_ERROR("Line %d: No pronunciation for word %s; ignored\n",
lineno, wptr[0]);
continue;
}
/* Convert pronunciation string to CI-phone-ids */
for (i = 1; i < nwd; i++) {
p[i - 1] = dict_ciphone_id(d, wptr[i]);
if (NOT_S3CIPID(p[i - 1])) {
E_ERROR("Line %d: Bad ciphone: %s; word %s ignored\n",
lineno, wptr[i], wptr[0]);
break;
}
}
if (i == nwd) { /* All CI-phones successfully converted to IDs */
w = dict_add_word(d, wptr[0], p, nwd - 1);
if (NOT_S3WID(w))
E_ERROR
("Line %d: dict_add_word (%s) failed (duplicate?); ignored\n",
lineno, wptr[0]);
else {
stralloc += strlen(d->word[w].word);
phnalloc += d->word[w].pronlen * sizeof(s3cipid_t);
}
}
}
E_INFO("Allocated %d KiB for strings, %d KiB for phones\n",
(int)stralloc / 1024, (int)phnalloc / 1024);
ckd_free(p);
ckd_free(wptr);
return 0;
}
static int32
dict_read(FILE * fp, dict_t * d)
{
char line[16384], **wptr;
s3cipid_t p[4096];
int32 lineno, nwd;
s3wid_t w;
int32 i, maxwd;
s3cipid_t ci;
int32 ph;
maxwd = 4092;
wptr = (char **) ckd_calloc(maxwd, sizeof(char *)); /* Freed below */
lineno = 0;
while (fgets(line, sizeof(line), fp) != NULL) {
lineno++;
if (line[0] == '#') /* Comment line */
continue;
if ((nwd = str2words(line, wptr, maxwd)) < 0)
E_FATAL("str2words(%s) failed; Increase maxwd from %d\n", line,
maxwd);
if (nwd == 0) /* Empty line */
continue;
/* wptr[0] is the word-string and wptr[1..nwd-1] the pronunciation sequence */
if (nwd == 1) {
E_ERROR("Line %d: No pronunciation for word %s; ignored\n",
lineno, wptr[0]);
continue;
}
{char * fin;
float proba=0.0;
int deca=0;
proba=strtod(wptr[1],&fin);
if (fin !=wptr[1])
deca=1;
else
proba=0.0;
/* Convert pronunciation string to CI-phone-ids */
for (i = 1; i < nwd-deca; i++) {
p[i - 1] = dict_ciphone_id(d, wptr[i+deca]);
if (NOT_S3CIPID(p[i - 1])) {
E_ERROR("Line %d: Bad ciphone: %s; word %s ignored\n",
lineno, wptr[i], wptr[0]);
break;
}
}
if (i == nwd-deca) { /* All CI-phones successfully converted to IDs */
w = dict_add_word(d, wptr[0], p, nwd - 1-deca);
if (NOT_S3WID(w))
E_ERROR
("Line %d: dict_add_word (%s) failed (duplicate?); ignored\n",
lineno, wptr[0]);
d->word[w].proba=proba;
}
}
}
if (d->lts_rules) {
#if 1 /* Until we allow user to put in a mapping of the phoneset from LTS to the phoneset from mdef,
The checking will intrusively stop the recognizer. */
for (ci = 0; ci < mdef_n_ciphone(d->mdef); ci++) {
if (!mdef_is_fillerphone(d->mdef, ci)) {
for (ph = 0; cmu6_lts_phone_table[ph] != NULL; ph++) {
/* E_INFO("%s %s\n",cmu6_lts_phone_table[ph],mdef_ciphone_str(d->mdef,ci)); */
if (!strcmp
(cmu6_lts_phone_table[ph],
mdef_ciphone_str(d->mdef, ci)))
break;
}
if (cmu6_lts_phone_table[ph] == NULL) {
E_FATAL
("A phone in the model definition doesn't appear in the letter to sound ",
"rules. \n This is case we don't recommend user to ",
"use the built-in LTS. \n Please kindly turn off ",
"-lts_mismatch\n");
}
}
}
#endif
}
