static void Open_JTalk_initialize(Open_JTalk * open_jtalk)
{
Mecab_initialize(&open_jtalk->mecab);
NJD_initialize(&open_jtalk->njd);
JPCommon_initialize(&open_jtalk->jpcommon);
HTS_Engine_initialize(&open_jtalk->engine);
}
void OpenJTalk_initialize(OpenJTalk * open_jtalk, HTS_Boolean use_lpf, int sampling_rate,
int fperiod, double alpha, int stage, double beta, int audio_buff_size,
double uv_threshold, HTS_Boolean use_log_gain, double gv_weight_mgc,
double gv_weight_lf0, double gv_weight_lpf)
{
Mecab_initialize(&open_jtalk->mecab);
NJD_initialize(&open_jtalk->njd);
JPCommon_initialize(&open_jtalk->jpcommon);
if (use_lpf)
HTS_Engine_initialize(&open_jtalk->engine, 3);
else
HTS_Engine_initialize(&open_jtalk->engine, 2);
HTS_Engine_set_sampling_rate(&open_jtalk->engine, sampling_rate);
HTS_Engine_set_fperiod(&open_jtalk->engine, fperiod);
HTS_Engine_set_alpha(&open_jtalk->engine, alpha);
HTS_Engine_set_gamma(&open_jtalk->engine, stage);
HTS_Engine_set_log_gain(&open_jtalk->engine, use_log_gain);
HTS_Engine_set_beta(&open_jtalk->engine, beta);
HTS_Engine_set_audio_buff_size(&open_jtalk->engine, audio_buff_size);
HTS_Engine_set_msd_threshold(&open_jtalk->engine, 1, uv_threshold);
HTS_Engine_set_gv_weight(&open_jtalk->engine, 0, gv_weight_mgc);
HTS_Engine_set_gv_weight(&open_jtalk->engine, 1, gv_weight_lf0);
if (use_lpf)
HTS_Engine_set_gv_weight(&open_jtalk->engine, 2, gv_weight_lpf);
}
static int setup(struct app *app)
{
#ifdef HTS_MELP
#define NR_STREAMS 3
#else
#define NR_STREAMS 2
#endif /* HTS_MELP */
double gv_weight[] = {
app->gv_weight_mgc,
app->gv_weight_lf0,
#ifdef HTS_MELP
app->gv_weight_lpf
#endif /* HTS_MELP */
};
int i;
app->play_h = play_init(&app->play_info, "default",
SND_PCM_FORMAT_S16_LE, 1, app->sampling_rate,
500000, 8);
Mecab_initialize(&app->mecab);
if (Mecab_load(&app->mecab, app->dn_mecab) != TRUE)
return -1;
NJD_initialize(&app->njd);
JPCommon_initialize(&app->jpcommon);
HTS_Engine_initialize(&app->engine);
if (HTS_Engine_load(&app->engine, &app->fn_voice, 1) != TRUE)
return -1;
HTS_Engine_set_sampling_frequency(&app->engine,
(size_t)app->sampling_rate);
if (app->fperiod >= 0)
HTS_Engine_set_fperiod(&app->engine, app->fperiod);
if (app->alpha >= 0.0)
HTS_Engine_set_alpha(&app->engine, app->alpha);
if (app->beta >= 0.0)
HTS_Engine_set_beta(&app->engine, app->beta);
if (app->half_tone >= 0.0)
HTS_Engine_add_half_tone(&app->engine, app->half_tone);
if (app->audio_buff_size > 0)
HTS_Engine_set_audio_buff_size(&app->engine,
app->audio_buff_size);
if (app->uv_threshold >= 0.0)
HTS_Engine_set_msd_threshold(&app->engine, 1,
app->uv_threshold);
if (app->speed >= 0.0)
HTS_Engine_set_speed(&app->engine, app->speed);
for (i = 0; i < NR_STREAMS; i++)
if (gv_weight[i] >= 0.0)
HTS_Engine_set_gv_weight(&app->engine, i, gv_weight[i]);
return 0;
}
void TextToSpeech::initialize(
int sampling_rate, int fperiod, double alpha, int stage, double beta, int audio_buff_size,
double uv_threshold, HTS_Boolean use_log_gain, double gv_weight_mgc,
double gv_weight_lf0, double gv_weight_lpf)
{
Mecab_initialize(&open_jtalk_.mecab);
NJD_initialize(&open_jtalk_.njd);
JPCommon_initialize(&open_jtalk_.jpcommon);
HTS_Engine_initialize(&open_jtalk_.engine, 2);
HTS_Engine_set_sampling_rate(&open_jtalk_.engine, sampling_rate);
HTS_Engine_set_fperiod(&open_jtalk_.engine, fperiod);
HTS_Engine_set_alpha(&open_jtalk_.engine, alpha);
HTS_Engine_set_gamma(&open_jtalk_.engine, stage);
HTS_Engine_set_log_gain(&open_jtalk_.engine, use_log_gain);
HTS_Engine_set_beta(&open_jtalk_.engine, beta);
HTS_Engine_set_audio_buff_size(&open_jtalk_.engine, audio_buff_size);
HTS_Engine_set_msd_threshold(&open_jtalk_.engine, 1, uv_threshold);
HTS_Engine_set_gv_weight(&open_jtalk_.engine, 0, gv_weight_mgc);
HTS_Engine_set_gv_weight(&open_jtalk_.engine, 1, gv_weight_lf0);
}
