/**
* Top function to read word dictionary via normal file pointer.
*
* @param fp [in] file pointer
* @param winfo [out] pointer to word dictionary to store the read data.
* @param hmminfo [in] HTK %HMM definition data. if NULL, phonemes are ignored.
* @param ignore_tri_conv [in] TRUE if triphone conversion is ignored
*
* @return TRUE on success, FALSE on any error word.
*/
boolean
voca_load_htkdict_fp(FILE *fp, WORD_INFO *winfo, HTK_HMM_INFO *hmminfo, boolean ignore_tri_conv)
{
boolean ret;
voca_load_start(winfo, hmminfo, ignore_tri_conv);
while(getl_fp(buf, MAXLINELEN, fp) != NULL) {
if (voca_load_line(buf, winfo, hmminfo) == FALSE) break;
}
ret = voca_load_end(winfo);
return(ret);
}
/**
* Top function to read word list via file pointer
*
* @param fp [in] file pointer
* @param winfo [out] pointer to word dictionary to store the read data.
* @param hmminfo [in] HTK %HMM definition data. if NULL, phonemes are ignored.
* @param headphone [in] word head silence model name
* @param tailphone [in] word tail silence model name
* @param contextphone [in] silence context name to be used at head and tail
*
* @return TRUE on success, FALSE on any error word.
*/
boolean
voca_load_wordlist_fp(FILE *fp, WORD_INFO *winfo, HTK_HMM_INFO *hmminfo, char *headphone, char *tailphone, char *contextphone)
{
boolean ret;
voca_load_start(winfo, hmminfo, FALSE);
while (getl_fp(buf, sizeof(buf), fp) != NULL) {
if (voca_load_word_line(buf, winfo, hmminfo, headphone, tailphone, contextphone) == FALSE) break;
}
ret = voca_load_end(winfo);
return(ret);
}
/**
* @brief Begin reading audio data from a file.
*
* If listfile was specified in adin_sndfile_standby(), the next filename
* will be read from the listfile. Otherwise, the
* filename will be obtained from stdin. Then the file will be opened here.
*
* @param filename [in] file name to open or NULL for prompt
*
* @return TRUE on success, FALSE on failure.
*/
boolean
adin_sndfile_begin(char *filename)
{
boolean readp;
if (filename != NULL) {
if (adin_sndfile_open(filename) == FALSE) {
jlog("Error: adin_sndfile: invalid format: \"%s\"\n", filename);
print_format(&sinfo);
return FALSE;
}
jlog("Stat: adin_sndfile: input speechfile: %s\n", filename);
print_format(&sinfo);
strcpy(speechfilename, filename);
return TRUE;
}
/* ready to read next input */
readp = FALSE;
while(readp == FALSE) {
if (from_file) {
/* read file name from listfile */
do {
if (getl_fp(speechfilename, MAXPATHLEN, fp_list) == NULL) { /* end of input */
fclose(fp_list);
return(FALSE); /* end of input */
}
} while (speechfilename[0] == '#'); /* skip comment */
} else {
/* read file name from stdin */
if (get_line_from_stdin(speechfilename, MAXPATHLEN, "enter filename->") == NULL) {
return (FALSE); /* end of input */
}
}
if (adin_sndfile_open(speechfilename) == FALSE) {
jlog("Error: adin_sndfile: invalid format: \"%s\"\n",speechfilename);
print_format(&sinfo);
} else {
jlog("Stat: adin_sndfile: input speechfile: %s\n",speechfilename);
print_format(&sinfo);
readp = TRUE;
}
}
return TRUE;
}
/**
* Top loop function to read DFA grammar via file descriptor
*
* @param fp [in] file pointer that points to the DFA grammar data
* @param dinfo [out] the read data will be stored in this DFA grammar structure
*
* @return TRUE on success, FALSE on failure.
*/
boolean
rddfa_fp(FILE *fp, DFA_INFO *dinfo)
{
int state_max, arc_num, terminal_max;
/* initialize */
dfa_state_init(dinfo);
state_max = 0;
arc_num = 0;
terminal_max = 0;
while(getl_fp(buf, MAXLINELEN, fp) != NULL) {
if (rddfa_line(buf, dinfo, &state_max, &arc_num, &terminal_max) == FALSE) {
break;
}
}
dinfo->state_num = state_max + 1;
dinfo->arc_num = arc_num;
dinfo->term_num = terminal_max + 1;
return(TRUE);
}
/**
* <JA>
* @brief 単語辞書をファイルから読み込んでセットアップする.
*
* 辞書上のモノフォン表記からトライフォンへの計算は init_voca() で
* 読み込み時に行われる. このため，辞書読み込み時には，認識で使用する
* 予定のHMM情報を与える必要がある.
*
* N-gram 使用時は，文頭無音単語およぶ文末無音単語をここで設定する.
* また，"-iwspword" 指定時は，ポーズ単語を辞書の最後に挿入する.
*
* </JA>
* <EN>
* @brief Read in word dictionary from a file and setup for recognition.
*
* Monophone-to-triphone conversion will be performed inside init_voca().
* So, an HMM definition data that will be used with the LM should also be
* specified as an argument.
*
* When reading dictionary for N-gram, sentence head silence word and
* tail silence word will be determined in this function. Also,
* when an option "-iwspword" is specified, this will insert a pause
* word at the last of the given dictionary.
*
* </EN>
*
* @param lmconf [in] LM configuration variables
* @param hmminfo [in] HMM definition of each phone in dictionary, for
* phone checking and monophone-to-triphone conversion.
*
* @return the newly created word dictionary structure, or NULL on failure.
*
*/
static WORD_INFO *
initialize_dict(JCONF_LM *lmconf, HTK_HMM_INFO *hmminfo)
{
WORD_INFO *winfo;
JCONF_LM_NAMELIST *nl;
char buf[MAXLINELEN];
int n;
/* allocate new word dictionary */
winfo = word_info_new();
/* read in dictinary from file */
if ( !
#ifdef MONOTREE
/* leave winfo monophone for 1st pass lexicon tree */
init_voca(winfo, lmconf->dictfilename, hmminfo, TRUE, lmconf->forcedict_flag)
#else
init_voca(winfo, lmconf->dictfilename, hmminfo, FALSE, lmconf->forcedict_flag)
#endif
) {
jlog("ERROR: m_fusion: failed to read dictionary, terminated\n");
word_info_free(winfo);
return NULL;
}
/* load additional entries */
for (nl = lmconf->additional_dict_files; nl; nl=nl->next) {
FILE *fp;
if ((fp = fopen(nl->name, "rb")) == NULL) {
jlog("ERROR: m_fusion: failed to open %s\n",nl->name);
word_info_free(winfo);
return NULL;
}
n = winfo->num;
while (getl_fp(buf, MAXLINELEN, fp) != NULL) {
if (voca_load_line(buf, winfo, hmminfo) == FALSE) break;
}
if (voca_load_end(winfo) == FALSE) {
if (lmconf->forcedict_flag) {
jlog("Warning: m_fusion: the error words above are ignored\n");
} else {
jlog("ERROR: m_fusion: error in reading dictionary %s\n", nl->name);
fclose(fp);
word_info_free(winfo);
return NULL;
}
}
if (fclose(fp) == -1) {
jlog("ERROR: m_fusion: failed to close %s\n", nl->name);
word_info_free(winfo);
return NULL;
}
jlog("STAT: + additional dictionary: %s (%d words)\n", nl->name, winfo->num - n);
}
n = winfo->num;
for (nl = lmconf->additional_dict_entries; nl; nl=nl->next) {
if (voca_load_line(nl->name, winfo, hmminfo) == FALSE) {
jlog("ERROR: m_fusion: failed to set entry: %s\n", nl->name);
}
}
if (lmconf->additional_dict_entries) {
if (voca_load_end(winfo) == FALSE) {
jlog("ERROR: m_fusion: failed to read additinoal word entry\n");
word_info_free(winfo);
return NULL;
}
jlog("STAT: + additional entries: %d words\n", winfo->num - n);
}
if (lmconf->lmtype == LM_PROB) {
/* if necessary, append a IW-sp word to the dict if "-iwspword" specified */
if (lmconf->enable_iwspword) {
if (
#ifdef MONOTREE
voca_append_htkdict(lmconf->iwspentry, winfo, hmminfo, TRUE)
#else
voca_append_htkdict(lmconf->iwspentry, winfo, hmminfo, FALSE)
#endif
== FALSE) {
jlog("ERROR: m_fusion: failed to make IW-sp word entry \"%s\"\n", lmconf->iwspentry);
word_info_free(winfo);
return NULL;
} else {
jlog("STAT: 1 IW-sp word entry added\n");
}
}
/* set {head,tail}_silwid */
winfo->head_silwid = voca_lookup_wid(lmconf->head_silname, winfo);
if (winfo->head_silwid == WORD_INVALID) { /* not exist */
jlog("ERROR: m_fusion: head sil word \"%s\" not exist in voca\n", lmconf->head_silname);
word_info_free(winfo);
return NULL;
}
winfo->tail_silwid = voca_lookup_wid(lmconf->tail_silname, winfo);
if (winfo->tail_silwid == WORD_INVALID) { /* not exist */
jlog("ERROR: m_fusion: tail sil word \"%s\" not exist in voca\n", lmconf->tail_silname);
word_info_free(winfo);
return NULL;
}
}
return(winfo);
}
/**
* Read and parse an HTK Config file, and set the specified option values.
*
* @param HTKconffile [in] HTK Config file path name
* @param para [out] MFCC parameter to set
*
* @return TRUE on success, FALSE on failure.
*/
boolean
htk_config_file_parse(char *HTKconffile, Value *para)
{
FILE *fp;
char buf[512];
char *p, *d, *a;
float srate;
boolean skipped;
jlog("Stat: para: parsing HTK Config file: %s\n", HTKconffile);
/* convert the content into argument list c_argv[1..c_argc-1] */
/* c_argv[0] will be the original conffile name */
if ((fp = fopen(HTKconffile, "r")) == NULL) {
jlog("Error: para: failed to open HTK Config file: %s\n", HTKconffile);
return FALSE;
}
srate = 0.0;
while (getl_fp(buf, 512, fp) != NULL) {
p = buf;
if (*p == 35) { /* skip comment line */
continue;
}
/* parse the input line to get directive and argument */
while (*p != '\0' && ISTOKEN(*p)) p++;
if (*p == '\0') continue;
d = p;
while (*p != '\0' && (!ISTOKEN(*p)) && *p != '=') p++;
if (*p == '\0') continue;
*p = '\0'; p++;
while (*p != '\0' && ((ISTOKEN(*p)) || *p == '=')) p++;
if (*p == '\0') continue;
a = p;
while (*p != '\0' && (!ISTOKEN(*p))) p++;
*p = '\0';
/* process arguments */
skipped = FALSE;
if (strmatch(d, "SOURCERATE")) { /* -smpPeriod */
srate = atof(a);
} else if (strmatch(d, "TARGETRATE")) { /* -fshift */
para->frameshift = atof(a);
} else if (strmatch(d, "WINDOWSIZE")) { /* -fsize */
para->framesize = atof(a);
} else if (strmatch(d, "ZMEANSOURCE")) { /* -zmeansource */
para->zmeanframe = (a[0] == 'T') ? TRUE : FALSE;
} else if (strmatch(d, "USEPOWER")) { /* -usepower */
para->usepower = (a[0] == 'T') ? TRUE : FALSE;
} else if (strmatch(d, "PREEMCOEF")) { /* -preemph */
para->preEmph = atof(a);
} else if (strmatch(d, "USEHAMMING")) { /* (fixed to T) */
if (a[0] != 'T') {
jlog("Error: para: USEHAMMING should be T\n", HTKconffile);
return FALSE;
}
} else if (strmatch(d, "NUMCHANS")) { /* -fbank */
para->fbank_num = atoi(a);
} else if (strmatch(d, "CEPLIFTER")) { /* -ceplif */
para->lifter = atoi(a);
} else if (strmatch(d, "DELTAWINDOW")) { /* -delwin */
para->delWin = atoi(a);
} else if (strmatch(d, "ACCWINDOW")) { /* -accwin */
para->accWin = atoi(a);
} else if (strmatch(d, "LOFREQ")) { /* -lofreq */
para->lopass = atof(a);
} else if (strmatch(d, "HIFREQ")) { /* -hifreq */
para->hipass = atof(a);
} else if (strmatch(d, "RAWENERGY")) { /* -rawe */
para->raw_e = (a[0] == 'T') ? TRUE : FALSE;
} else if (strmatch(d, "ENORMALISE")) { /* -enormal */
para->enormal = (a[0] == 'T') ? TRUE : FALSE;
} else if (strmatch(d, "ESCALE")) { /* -escale */
para->escale = atof(a);
} else if (strmatch(d, "SILFLOOR")) { /* -silfloor */
para->silFloor = atof(a);
} else if (strmatch(d, "WARPFREQ")) { /* -vtln (1) */
para->vtln_alpha = atof(a);
} else if (strmatch(d, "WARPLCUTOFF")) { /* -vtln (2) */
para->vtln_lower = atof(a);
} else if (strmatch(d, "WARPUCUTOFF")) { /* -vtln (3) */
para->vtln_upper = atof(a);
} else if (strmatch(d, "TARGETKIND")) {
jlog("Warning: para: TARGETKIND skipped (will be determined by AM header)\n");
skipped = TRUE;
} else if (strmatch(d, "NUMCEPS")) {
jlog("Warning: para: NUMCEPS skipped (will be determined by AM header)\n");
skipped = TRUE;
} else {
jlog("Warning: para: \"%s\" ignored (not supported, or irrelevant)\n", d);
skipped = TRUE;
}
if (!skipped) {
jlog("Stat: para: %s=%s\n", d, a);
}
}
if (srate == 0.0) {
jlog("Warning: no SOURCERATE found\n");
jlog("Warning: assume source waveform sample rate to 625 (16kHz)\n");
srate = 625;
}
para->smp_period = srate;
para->smp_freq = period2freq(para->smp_period);
para->frameshift /= srate;
para->framesize /= srate;
if (fclose(fp) == -1) {
jlog("Error: para: failed to close file\n");
return FALSE;
}
para->loaded = 1;
return TRUE;
}
