float flite_file_to_speech(const char *filename,
cst_voice *voice,
const char *outtype)
{
cst_utterance *utt;
cst_tokenstream *ts;
const char *token;
cst_item *t;
cst_relation *tokrel;
float durs = 0;
int num_tokens;
cst_wave *w;
cst_breakfunc breakfunc = default_utt_break;
cst_uttfunc utt_user_callback = 0;
int fp;
if ((ts = ts_open(filename,
get_param_string(voice->features,"text_whitespace",NULL),
get_param_string(voice->features,"text_singlecharsymbols",NULL),
get_param_string(voice->features,"text_prepunctuation",NULL),
get_param_string(voice->features,"text_postpunctuation",NULL)))
== NULL)
{
cst_errmsg("failed to open file \"%s\" for reading\n",
filename);
return 1;
}
fp = get_param_int(voice->features,"file_start_position",0);
if (fp > 0)
ts_set_stream_pos(ts,fp);
if (feat_present(voice->features,"utt_break"))
breakfunc = val_breakfunc(feat_val(voice->features,"utt_break"));
if (feat_present(voice->features,"utt_user_callback"))
utt_user_callback = val_uttfunc(feat_val(voice->features,"utt_user_callback"));
/* If its a file to write to, create and save an empty wave file */
/* as we are going to incrementally append to it */
if (!cst_streq(outtype,"play") &&
!cst_streq(outtype,"none") &&
!cst_streq(outtype,"stream"))
{
w = new_wave();
cst_wave_resize(w,0,1);
cst_wave_set_sample_rate(w,16000);
cst_wave_save_riff(w,outtype); /* an empty wave */
delete_wave(w);
}
num_tokens = 0;
utt = new_utterance();
tokrel = utt_relation_create(utt, "Token");
while (!ts_eof(ts) || num_tokens > 0)
{
token = ts_get(ts);
if ((cst_strlen(token) == 0) ||
(num_tokens > 500) || /* need an upper bound */
(relation_head(tokrel) &&
breakfunc(ts,token,tokrel)))
{
/* An end of utt, so synthesize it */
if (utt_user_callback)
utt = (utt_user_callback)(utt);
if (utt)
{
utt = flite_do_synth(utt,voice,utt_synth_tokens);
durs += flite_process_output(utt,outtype,TRUE);
delete_utterance(utt); utt = NULL;
}
else
break;
if (ts_eof(ts)) break;
utt = new_utterance();
tokrel = utt_relation_create(utt, "Token");
num_tokens = 0;
}
num_tokens++;
t = relation_append(tokrel, NULL);
item_set_string(t,"name",token);
item_set_string(t,"whitespace",ts->whitespace);
item_set_string(t,"prepunctuation",ts->prepunctuation);
item_set_string(t,"punc",ts->postpunctuation);
/* Mark it at the beginning of the token */
item_set_int(t,"file_pos",
ts->file_pos-(1+ /* as we are already on the next char */
cst_strlen(token)+
cst_strlen(ts->prepunctuation)+
cst_strlen(ts->postpunctuation)));
item_set_int(t,"line_number",ts->line_number);
}
delete_utterance(utt);
ts_close(ts);
return durs;
}
int main(int argc, char **argv)
{
cst_track *t1;
cst_track *me_filters = NULL;
cst_wave *w1, *w2, *res = NULL;
cst_val *files;
cst_features *args;
int i, j;
int order, o, s;
int frame_length;
float *lpcs, *residual;
float m;
const char *f1, *f2;
const char *resfn = NULL;
int last_peak = 0, next_peak;
int period;
float power;
int rfc = 0;
int str = 0;
int fn, fo, ss;
float xpulse, xnoise;
float fxpulse, fxnoise;
float x, me;
float *hpulse = NULL, *hnoise = NULL;
float *xpulsesig = NULL, *xnoisesig = NULL;
int q = 0;
int position;
int lpc_start = 0;
args = new_features();
files =
cst_args(argv, argc,
"usage: lpc_resynth OPTIONS INTRACK OUTWAVE\n"
"Resynth an lpc track\n"
"-res <string> residual (as waveform)\n"
"-save_res Save the generated residual\n"
"-lpc_start <int> start of lpc params in lpc track {1}\n"
"-order <int> LPC order {16}\n"
"-str mixed excitation strengths\n"
"-me_filters <string> mixed excitation filters\n"
"-rfc Coefficents are reflection coefficients\n", args);
f1 = val_string(val_car(files));
f2 = val_string(val_car(val_cdr(files)));
t1 = new_track();
lpc_start = mimic_get_param_int(args, "-lpc_start", 1);
if (feat_present(args, "-rfc"))
rfc = 1;
if (feat_present(args, "-str"))
str = 1;
if (feat_present(args, "-me_filters"))
{
me_filters = new_track();
if (cst_track_load_est
(me_filters,
mimic_get_param_string(args, "-me_filters",
"me_filters.track")) != CST_OK_FORMAT)
{
fprintf(stderr,
"lpc_resynth: can't read file or wrong format \"%s\"\n",
f1);
return -1;
}
hpulse = cst_alloc(float, me_filters->num_channels);
hnoise = cst_alloc(float, me_filters->num_channels);
xpulsesig = cst_alloc(float, me_filters->num_channels);
xnoisesig = cst_alloc(float, me_filters->num_channels);
}
int main(int argc, char **argv)
{
bard_reader *br;
cst_features *args;
cst_features *config;
float idle_time;
int spos;
args = new_features();
cst_args(argv,argc,
"usage: bard OPTIONS\n"
"Bard Storyteller\n"
"-text <string> Text filename to read\n"
"-text_pos <int> Starting position in text file\n"
"-config_file <string> Override default config file ($HOME/.bard_config)\n"
"-font <string> Pathname to ttf font file\n"
"-font_size <int> Initial font size\n"
"-audio_method <string> SDL or flite\n"
"-voice <string> Voice name (or pathname to dumped voice)\n"
"-audio_stream_buffer_factor <int> Bigger for slower processors\n"
"-gain <float> Volume factor (1.0 is default)\n"
"-speed <float> Duration stretch (inverse speed)\n"
"-battery_script <string> Script to get battery charge state\n"
"-screen_height <int> \n"
"-screen_width <int> \n"
"-screen_blank_idle_time <float> In secs (0 means no blanking)\n"
"-scroll_delay <int> In ms\n"
"-voices_dir <string> Directory containing *.flitevox voices\n"
"-debug <int> Print debug messages\n"
"",
args);
/* If text specified and no position given, start from 0 */
if (feat_present(args,"-text") && !feat_present(args,"-text_pos"))
feat_set_int(args,"-text_pos",0);
if (!feat_present(args,"-font"))
feat_set_string(args,"-font",BARD_DEFAULT_FONT);
bard_debug = get_param_int(args,"-debug",0);
config = bard_read_config(args);
cst_feat_print(stdout,config);
br = bard_open(config);
if (br == NULL) return -1; /* something went really wrong */
/* Make the windows */
br->text = bard_make_text_window(br);
br->file_select = bard_make_file_select_window(br);
br->info = bard_make_info_window(br); /* general info/params */
br->help = bard_make_help_window(br);
br->recent = bard_make_recent_window(br);
br->menu = bard_make_menu_window(br);
if (feat_present(br->config,"-voices_dir"))
br->voice_select = bard_make_voice_select_window(br);
/* Select focus */
br->display->current = br->text;
bard_display_clear(br->display);
bard_display_update(br->display);
bard_screen_on(NULL); /* just in case it is currently off */
while (br->quit == 0)
{
if (input_process_events(br) == 0)
{
if (br->scroll)
{
SDL_Delay(br->scroll_delay); /* no key press but scrolling */
br->quiet_time = SDL_GetTicks(); /* no blank while scrolling */
}
else
SDL_Delay(br->no_key_delay); /* no key press, so pause a bit */
}
if (br->speak == 1)
{
/* Within the audio callback function, we also call */
/* input_process_events so we can stop if requested */
br->pause = 0;
if (br->display->current->current_token)
spos = br->display->current->current_token->file_pos;
else
spos = br->display->current->sop_pos;
/* We need spos-1 here to make it display well */
bard_speak_text(br->speech,br->display->current->textfile,spos-1);
br->speak = 0;
br->quiet_time = SDL_GetTicks();
}
else if (br->scroll)
{ /* Speaking and smooth scrolling are mutually exclusive */
if (bard_text_scroll(br->text) == 0)
br->scroll = 0; /* Reached end of file so stop scroll */
bard_display_update(br->display); /* Update screen */
}
else
br->text->scroll_offset = 0;
/* Check if we should black/hide cursor */
idle_time = (SDL_GetTicks()-br->quiet_time)/1000.0;
if ((br->screen_blank_idle_time > 0) &&
(idle_time > br->screen_blank_idle_time) &&
(br->blank == 0))
{
br->blank = 1;
bard_screen_off(NULL);
}
idle_time = (SDL_GetTicks()-br->mouse_quiet_time)/1000.0;
if (idle_time > 5.0)
SDL_ShowCursor(SDL_DISABLE);
}
/* Save context */
bard_write_config(br);
bard_close(br);
delete_features(args);
return 0;
}
