sumom****@users*****
sumom****@users*****
2013年 12月 18日 (水) 12:55:21 JST
Index: julius4/libsent/src/hmminfo/put_htkdata_info.c
diff -u julius4/libsent/src/hmminfo/put_htkdata_info.c:1.8 julius4/libsent/src/hmminfo/put_htkdata_info.c:1.9
--- julius4/libsent/src/hmminfo/put_htkdata_info.c:1.8 Fri Jun 21 02:14:21 2013
+++ julius4/libsent/src/hmminfo/put_htkdata_info.c Wed Dec 18 12:55:21 2013
@@ -12,7 +12,7 @@
* @author Akinobu LEE
* @date Tue Feb 15 23:36:00 2005
*
- * $Revision: 1.8 $
+ * $Revision: 1.9 $
*
*/
/*
@@ -148,6 +148,7 @@
fprintf(fp, "no output state\n");
} else {
if (s->name != NULL) fprintf(fp, "[~s \"%s\"]\n", s->name);
+ fprintf(fp, "id: %d\n", s->id);
for (st=0;st<s->nstream;st++) {
fprintf(fp, "stream %d:", st + 1);
if (s->w != NULL) {
Index: julius4/libsent/src/hmminfo/rdhmmdef.c
diff -u julius4/libsent/src/hmminfo/rdhmmdef.c:1.8 julius4/libsent/src/hmminfo/rdhmmdef.c:1.9
--- julius4/libsent/src/hmminfo/rdhmmdef.c:1.8 Fri Jun 21 02:14:21 2013
+++ julius4/libsent/src/hmminfo/rdhmmdef.c Wed Dec 18 12:55:21 2013
@@ -30,7 +30,7 @@
* @author Akinobu LEE
* @date Wed Feb 16 00:17:18 2005
*
- * $Revision: 1.8 $
+ * $Revision: 1.9 $
*
*/
/*
@@ -43,10 +43,14 @@
#include <sent/stddefs.h>
#include <sent/htk_param.h>
#include <sent/htk_hmm.h>
+#ifdef HAVE_ZLIB
+#include <zlib.h>
+#endif
#define MAXBUFLEN 4096 ///< Maximum length of a line in the input
char *rdhmmdef_token; ///< Current token string (GLOBAL)
+static boolean last_line_full = FALSE;
static char buf[MAXBUFLEN]; ///< Local work area for token reading
static int line; ///< Input Line count
@@ -79,17 +83,52 @@
char *
read_token(FILE *fp)
{
+ int len;
+ int bp = 0;
+ int maxlen = MAXBUFLEN;
+ static char delims[] = HMMDEF_DELM;
+
if ((rdhmmdef_token = mystrtok_quote(NULL, HMMDEF_DELM)) != NULL) {
- /* return next token */
- return rdhmmdef_token;
+ /* has token */
+ if (mystrtok_movetonext(NULL, HMMDEF_DELM) != NULL || last_line_full == FALSE) {
+ /* return the current token, if this is not a last token, or
+ last is newline terminated */
+ return rdhmmdef_token;
+ } else {
+ /* concatinate the last token with next line */
+ len = strlen(rdhmmdef_token);
+ memmove(buf, rdhmmdef_token, len);
+ bp = len;
+ maxlen -= len;
+ }
}
- /* read new 1 line */
- if (getl(buf, MAXBUFLEN, fp) == NULL) {
- rdhmmdef_token = NULL;
- } else {
- rdhmmdef_token = mystrtok_quote(buf, HMMDEF_DELM);
- line++;
+
+ /* read new 1 line a*/
+ while(
+#ifdef HAVE_ZLIB
+ gzgets((gzFile)fp, &(buf[bp]), maxlen) != Z_NULL
+#else
+ fgets(&(buf[bp]), maxlen, fp) != NULL
+#endif
+ ) {
+ /* chop delimiters at end of line (incl. newline) */
+ /* if no delimiter at end of line, last_line_full is TRUE */
+ last_line_full = TRUE;
+ len = strlen(buf)-1;
+ while (len >= 0 && strchr(delims, buf[len])) {
+ last_line_full = FALSE;
+ buf[len--] = '\0';
+ }
+ if (buf[0] != '\0') {
+ /* start getting next token */
+ rdhmmdef_token = mystrtok_quote(buf, HMMDEF_DELM);
+ /* increment line */
+ line++;
+ return rdhmmdef_token;
+ }
}
+ /* when reading error, return NULL */
+ rdhmmdef_token = NULL;
return rdhmmdef_token;
}
@@ -162,6 +201,52 @@
}
#endif
+boolean
+htk_hmm_check_sid(HTK_HMM_INFO *hmm)
+{
+ HTK_HMM_State *stmp;
+ boolean *check;
+ int i;
+ boolean ok_p;
+
+ /* check if each state is assigned a valid sid */
+ check = (boolean *)mymalloc(sizeof(boolean) * hmm->totalstatenum);
+ for(i = 0; i < hmm->totalstatenum; i++) check[i] = FALSE;
+ for (stmp = hmm->ststart; stmp; stmp = stmp->next) {
+ if (stmp->id == -1) {
+ jlog("Error: rdhmmdef: no SID on some states\n");
+ free(check);
+ return FALSE;
+ }
+ if (stmp->id < 0) {
+ jlog("Error: rdhmmdef: invalid SID value: %d\n", stmp->id);
+ free(check);
+ return FALSE;
+ }
+ if (stmp->id >= hmm->totalstatenum) {
+ jlog("Error: rdhmmdef: SID value exceeds the number of states? (%d > %d)\n", stmp->id, hmm->totalstatenum);
+ free(check);
+ return FALSE;
+ }
+ if (check[stmp->id] == TRUE) {
+ jlog("Error: rdhmmdef: duplicate definition to the same SID: %d\n", stmp->id);
+ free(check);
+ return FALSE;
+ }
+ check[stmp->id] = TRUE;
+ }
+ ok_p = TRUE;
+ for(i = 0; i < hmm->totalstatenum; i++) {
+ if (check[i] == FALSE) {
+ jlog("Error: rdhmmdef: missing SID: %d\n", i);
+ ok_p = FALSE;
+ }
+ }
+ free(check);
+
+ return ok_p;
+}
+
/**
* @brief Main top routine to read in HTK %HMM definition file.
*
@@ -288,25 +373,54 @@
return FALSE;
}
- /* add ID number for all HTK_HMM_State */
- /* also calculate the maximum number of mixture */
+ /* add ID number for all HTK_HMM_State if not assigned */
{
HTK_HMM_State *stmp;
- int n, max, s, mix;
+ int n;
+ boolean has_sid;
+
+ /* caclculate total num and check if has sid */
+ has_sid = FALSE;
n = 0;
- max = 0;
for (stmp = hmm->ststart; stmp; stmp = stmp->next) {
- for(s=0;s<stmp->nstream;s++) {
- mix = stmp->pdf[s]->mix_num;
- if (max < mix) max = mix;
- }
- stmp->id = n++;
+ n++;
if (n >= MAX_STATE_NUM) {
jlog("Error: rdhmmdef: too much states in a model > %d\n", MAX_STATE_NUM);
return FALSE;
}
+ if (stmp->id != -1) {
+ has_sid = TRUE;
+ }
}
hmm->totalstatenum = n;
+ if (has_sid) {
+ jlog("Stat: rdhmmdef: <SID> found in the definition\n");
+ /* check if each state is assigned a valid sid */
+ if (htk_hmm_check_sid(hmm) == FALSE) {
+ jlog("Error: rdhmmdef: error in SID\n");
+ return FALSE;
+ }
+ } else {
+ /* assign internal sid (will not be saved) */
+ jlog("Stat: rdhmmdef: no <SID> embedded\n");
+ jlog("Stat: rdhmmdef: assign SID by the order of appearance\n");
+ n = hmm->totalstatenum;
+ for (stmp = hmm->ststart; stmp; stmp = stmp->next) {
+ stmp->id = --n;
+ }
+ }
+ }
+ /* calculate the maximum number of mixture */
+ {
+ HTK_HMM_State *stmp;
+ int max, s, mix;
+ max = 0;
+ for (stmp = hmm->ststart; stmp; stmp = stmp->next) {
+ for(s=0;s<stmp->nstream;s++) {
+ mix = stmp->pdf[s]->mix_num;
+ if (max < mix) max = mix;
+ }
+ }
hmm->maxmixturenum = max;
}
/* compute total number of HMM models and maximum length */
Index: julius4/libsent/src/hmminfo/rdhmmdef_state.c
diff -u julius4/libsent/src/hmminfo/rdhmmdef_state.c:1.7 julius4/libsent/src/hmminfo/rdhmmdef_state.c:1.8
--- julius4/libsent/src/hmminfo/rdhmmdef_state.c:1.7 Fri Jun 21 02:14:21 2013
+++ julius4/libsent/src/hmminfo/rdhmmdef_state.c Wed Dec 18 12:55:21 2013
@@ -12,7 +12,7 @@
* @author Akinobu LEE
* @date Wed Feb 16 03:07:44 2005
*
- * $Revision: 1.7 $
+ * $Revision: 1.8 $
*
*/
/*
@@ -46,7 +46,7 @@
for(i=0;i<new->nstream;i++) {
new->pdf[i] = NULL;
}
- new->id = 0;
+ new->id = -1;
new->next = NULL;
return(new);
@@ -128,6 +128,13 @@
new = state_new(hmm);
+ if (currentis("SID")) {
+ read_token(fp);
+ NoTokErr("missing SID value");
+ new->id = atoi(rdhmmdef_token);
+ read_token(fp);
+ }
+
if (currentis("NUMMIXES")) {
if (hmm->tmp_mixnum == NULL) {
hmm->tmp_mixnum = (int *)mybmalloc2(sizeof(int) * hmm->opt.stream_info.num, &(hmm->mroot));
@@ -188,7 +195,7 @@
{
HTK_HMM_State *tmp;
- if (currentis("NUMMIXES")||currentis("SWEIGHTS")||currentis("~w")||currentis("STREAM")||currentis("MIXTURE")||currentis("TMIX")||currentis("MEAN")||currentis("~m")||currentis("RCLASS")) {
+ if (currentis("SID")||currentis("NUMMIXES")||currentis("SWEIGHTS")||currentis("~w")||currentis("STREAM")||currentis("MIXTURE")||currentis("TMIX")||currentis("MEAN")||currentis("~m")||currentis("RCLASS")) {
/* definition: define state data, and return the pointer */
tmp = state_read(fp, hmm);
tmp->name = NULL; /* no name */
Index: julius4/libsent/src/hmminfo/read_binhmm.c
diff -u julius4/libsent/src/hmminfo/read_binhmm.c:1.10 julius4/libsent/src/hmminfo/read_binhmm.c:1.11
--- julius4/libsent/src/hmminfo/read_binhmm.c:1.10 Fri Jun 21 02:14:21 2013
+++ julius4/libsent/src/hmminfo/read_binhmm.c Wed Dec 18 12:55:21 2013
@@ -22,7 +22,7 @@
* @author Akinobu LEE
* @date Wed Feb 16 05:23:59 2005
*
- * $Revision: 1.10 $
+ * $Revision: 1.11 $
*
*/
/*
@@ -862,13 +862,22 @@
hmm->totalpdfnum = n;
}
- /* re-number state id */
+ /* check state id */
{
HTK_HMM_State *stmp;
- int n = 0;
+ int n;
+ boolean has_sid;
+
+ /* check if each state is assigned a valid sid */
+ if (htk_hmm_check_sid(hmm) == FALSE) {
+ jlog("Error: rdhmmdef: error in SID\n");
+ return FALSE;
+ }
+#if 0
for (stmp = hmm->ststart; stmp; stmp = stmp->next) {
stmp->id = n++;
}
+#endif
}
/* assign ID number for all HTK_HMM_Trans */
{
Index: julius4/libsent/src/hmminfo/write_binhmm.c
diff -u julius4/libsent/src/hmminfo/write_binhmm.c:1.8 julius4/libsent/src/hmminfo/write_binhmm.c:1.9
--- julius4/libsent/src/hmminfo/write_binhmm.c:1.8 Fri Jun 21 02:14:21 2013
+++ julius4/libsent/src/hmminfo/write_binhmm.c Wed Dec 18 12:55:21 2013
@@ -22,7 +22,7 @@
* @author Akinobu LEE
* @date Wed Feb 16 06:03:36 2005
*
- * $Revision: 1.8 $
+ * $Revision: 1.9 $
*
*/
/*
@@ -31,7 +31,7 @@
* All rights reserved
*/
-/* $Id: write_binhmm.c,v 1.8 2013/06/20 17:14:21 sumomo Exp $ */
+/* $Id: write_binhmm.c,v 1.9 2013/12/18 03:55:21 sumomo Exp $ */
#include <sent/stddefs.h>
#include <sent/htk_param.h>
@@ -823,8 +823,9 @@
static int
qsort_st_index(HTK_HMM_State **s1, HTK_HMM_State **s2)
{
- if (*s1 > *s2) return 1;
- else if (*s1 < *s2) return -1;
+ /* keep ID order */
+ if ((*s1)->id > (*s2)->id) return 1;
+ else if ((*s1)->id < (*s2)->id) return -1;
else return 0;
}
@@ -918,7 +919,8 @@
while (left < right) {
mid = (left + right) / 2;
- if (st_index[mid] < s) {
+ /* search by id */
+ if (st_index[mid]->id < s->id) {
left = mid + 1;
} else {
right = mid;