eSpeak NG is an open source speech synthesizer that supports more than hundred languages and accents.
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

speak_lib.cpp 32KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300
  1. /***************************************************************************
  2. * Copyright (C) 2005 to 2013 by Jonathan Duddington *
  3. * email: [email protected] *
  4. * Copyright (C) 2013 Reece H. Dunn *
  5. * *
  6. * This program is free software; you can redistribute it and/or modify *
  7. * it under the terms of the GNU General Public License as published by *
  8. * the Free Software Foundation; either version 3 of the License, or *
  9. * (at your option) any later version. *
  10. * *
  11. * This program is distributed in the hope that it will be useful, *
  12. * but WITHOUT ANY WARRANTY; without even the implied warranty of *
  13. * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the *
  14. * GNU General Public License for more details. *
  15. * *
  16. * You should have received a copy of the GNU General Public License *
  17. * along with this program; if not, write see: *
  18. * <http://www.gnu.org/licenses/>. *
  19. ***************************************************************************/
  20. #include "StdAfx.h"
  21. #include "stdio.h"
  22. #include "ctype.h"
  23. #include "string.h"
  24. #include "stdlib.h"
  25. #include "wchar.h"
  26. #include "locale.h"
  27. #include <assert.h>
  28. #include <time.h>
  29. #include "speech.h"
  30. #include <sys/stat.h>
  31. #ifdef PLATFORM_WINDOWS
  32. #include <fcntl.h>
  33. #include <io.h>
  34. #include <windows.h>
  35. #include <winreg.h>
  36. #else /* PLATFORM_POSIX */
  37. #include <unistd.h>
  38. #endif
  39. #include "speak_lib.h"
  40. #include "phoneme.h"
  41. #include "synthesize.h"
  42. #include "voice.h"
  43. #include "translate.h"
  44. #include "debug.h"
  45. #include "fifo.h"
  46. #include "event.h"
  47. #include "wave.h"
  48. unsigned char *outbuf=NULL;
  49. espeak_EVENT *event_list=NULL;
  50. int event_list_ix=0;
  51. int n_event_list;
  52. long count_samples;
  53. void* my_audio=NULL;
  54. static unsigned int my_unique_identifier=0;
  55. static void* my_user_data=NULL;
  56. static espeak_AUDIO_OUTPUT my_mode=AUDIO_OUTPUT_SYNCHRONOUS;
  57. static int synchronous_mode = 1;
  58. static int out_samplerate = 0;
  59. static int voice_samplerate = 22050;
  60. static espeak_ERROR err = EE_OK;
  61. t_espeak_callback* synth_callback = NULL;
  62. int (* uri_callback)(int, const char *, const char *) = NULL;
  63. int (* phoneme_callback)(const char *) = NULL;
  64. char path_home[N_PATH_HOME]; // this is the espeak-data directory
  65. extern int saved_parameters[N_SPEECH_PARAM]; //Parameters saved on synthesis start
  66. void WVoiceChanged(voice_t *wvoice)
  67. {//=================================
  68. // Voice change in wavegen
  69. voice_samplerate = wvoice->samplerate;
  70. }
  71. #ifdef USE_ASYNC
  72. static int dispatch_audio(short* outbuf, int length, espeak_EVENT* event)
  73. {//======================================================================
  74. ENTER("dispatch_audio");
  75. int a_wave_can_be_played = fifo_is_command_enabled();
  76. #ifdef DEBUG_ENABLED
  77. SHOW("*** dispatch_audio > uid=%d, [write=%p (%d bytes)], sample=%d, a_wave_can_be_played = %d\n",
  78. (event) ? event->unique_identifier : 0, wave_test_get_write_buffer(), 2*length,
  79. (event) ? event->sample : 0,
  80. a_wave_can_be_played);
  81. #endif
  82. switch(my_mode)
  83. {
  84. case AUDIO_OUTPUT_PLAYBACK:
  85. {
  86. int event_type=0;
  87. if(event)
  88. {
  89. event_type = event->type;
  90. }
  91. if(event_type == espeakEVENT_SAMPLERATE)
  92. {
  93. voice_samplerate = event->id.number;
  94. if(out_samplerate != voice_samplerate)
  95. {
  96. if(out_samplerate != 0)
  97. {
  98. // sound was previously open with a different sample rate
  99. wave_close(my_audio);
  100. sleep(1);
  101. }
  102. out_samplerate = voice_samplerate;
  103. if(!wave_init(voice_samplerate))
  104. {
  105. err = EE_INTERNAL_ERROR;
  106. return(-1);
  107. }
  108. wave_set_callback_is_output_enabled( fifo_is_command_enabled);
  109. my_audio = wave_open("alsa");
  110. event_init();
  111. }
  112. }
  113. if (outbuf && length && a_wave_can_be_played)
  114. {
  115. wave_write (my_audio, (char*)outbuf, 2*length);
  116. }
  117. while(a_wave_can_be_played) {
  118. // TBD: some event are filtered here but some insight might be given
  119. // TBD: in synthesise.cpp for avoiding to create WORDs with size=0.
  120. // TBD: For example sentence "or ALT)." returns three words
  121. // "or", "ALT" and "".
  122. // TBD: the last one has its size=0.
  123. if (event && (event->type == espeakEVENT_WORD) && (event->length==0))
  124. {
  125. break;
  126. }
  127. espeak_ERROR a_error = event_declare(event);
  128. if (a_error != EE_BUFFER_FULL)
  129. {
  130. break;
  131. }
  132. SHOW_TIME("dispatch_audio > EE_BUFFER_FULL\n");
  133. usleep(10000);
  134. a_wave_can_be_played = fifo_is_command_enabled();
  135. }
  136. }
  137. break;
  138. case AUDIO_OUTPUT_RETRIEVAL:
  139. if (synth_callback)
  140. {
  141. synth_callback(outbuf, length, event);
  142. }
  143. break;
  144. case AUDIO_OUTPUT_SYNCHRONOUS:
  145. case AUDIO_OUTPUT_SYNCH_PLAYBACK:
  146. break;
  147. }
  148. if (!a_wave_can_be_played)
  149. {
  150. SHOW_TIME("dispatch_audio > synth must be stopped!\n");
  151. }
  152. SHOW_TIME("LEAVE dispatch_audio\n");
  153. return (a_wave_can_be_played==0); // 1 = stop synthesis, -1 = error
  154. }
  155. static int create_events(short* outbuf, int length, espeak_EVENT* event, uint32_t the_write_pos)
  156. {//=====================================================================
  157. int finished;
  158. int i=0;
  159. // The audio data are written to the output device.
  160. // The list of events in event_list (index: event_list_ix) is read:
  161. // Each event is declared to the "event" object which stores them internally.
  162. // The event object is responsible of calling the external callback
  163. // as soon as the relevant audio sample is played.
  164. do
  165. { // for each event
  166. espeak_EVENT* event;
  167. if (event_list_ix == 0)
  168. {
  169. event = NULL;
  170. }
  171. else
  172. {
  173. event = event_list + i;
  174. #ifdef DEBUG_ENABLED
  175. SHOW("Synthesize: event->sample(%d) + %d = %d\n", event->sample, the_write_pos, event->sample + the_write_pos);
  176. #endif
  177. event->sample += the_write_pos;
  178. }
  179. #ifdef DEBUG_ENABLED
  180. SHOW("*** Synthesize: i=%d (event_list_ix=%d), length=%d\n",i,event_list_ix,length);
  181. #endif
  182. finished = dispatch_audio((short *)outbuf, length, event);
  183. length = 0; // the wave data are played once.
  184. i++;
  185. } while((i < event_list_ix) && !finished);
  186. return finished;
  187. }
  188. int sync_espeak_terminated_msg( uint32_t unique_identifier, void* user_data)
  189. {//=====================================================================
  190. ENTER("sync_espeak_terminated_msg");
  191. int finished=0;
  192. memset(event_list, 0, 2*sizeof(espeak_EVENT));
  193. event_list[0].type = espeakEVENT_MSG_TERMINATED;
  194. event_list[0].unique_identifier = unique_identifier;
  195. event_list[0].user_data = user_data;
  196. event_list[1].type = espeakEVENT_LIST_TERMINATED;
  197. event_list[1].unique_identifier = unique_identifier;
  198. event_list[1].user_data = user_data;
  199. if (my_mode==AUDIO_OUTPUT_PLAYBACK)
  200. {
  201. while(1)
  202. {
  203. espeak_ERROR a_error = event_declare(event_list);
  204. if (a_error != EE_BUFFER_FULL)
  205. {
  206. break;
  207. }
  208. SHOW_TIME("sync_espeak_terminated_msg > EE_BUFFER_FULL\n");
  209. usleep(10000);
  210. }
  211. }
  212. else
  213. {
  214. if (synth_callback)
  215. {
  216. finished=synth_callback(NULL,0,event_list);
  217. }
  218. }
  219. return finished;
  220. }
  221. #endif
  222. static void select_output(espeak_AUDIO_OUTPUT output_type)
  223. {//=======================================================
  224. my_mode = output_type;
  225. my_audio = NULL;
  226. synchronous_mode = 1;
  227. option_waveout = 1; // inhibit portaudio callback from wavegen.cpp
  228. out_samplerate = 0;
  229. switch(my_mode)
  230. {
  231. case AUDIO_OUTPUT_PLAYBACK:
  232. // wave_init() is now called just before the first wave_write()
  233. synchronous_mode = 0;
  234. break;
  235. case AUDIO_OUTPUT_RETRIEVAL:
  236. synchronous_mode = 0;
  237. break;
  238. case AUDIO_OUTPUT_SYNCHRONOUS:
  239. break;
  240. case AUDIO_OUTPUT_SYNCH_PLAYBACK:
  241. option_waveout = 0;
  242. WavegenInitSound();
  243. break;
  244. }
  245. } // end of select_output
  246. int GetFileLength(const char *filename)
  247. {//====================================
  248. struct stat statbuf;
  249. if(stat(filename,&statbuf) != 0)
  250. return(0);
  251. if((statbuf.st_mode & S_IFMT) == S_IFDIR)
  252. // if(S_ISDIR(statbuf.st_mode))
  253. return(-2); // a directory
  254. return(statbuf.st_size);
  255. } // end of GetFileLength
  256. char *Alloc(int size)
  257. {//==================
  258. char *p;
  259. if((p = (char *)malloc(size)) == NULL)
  260. fprintf(stderr,"Can't allocate memory\n"); // I was told that size+1 fixes a crash on 64-bit systems
  261. return(p);
  262. }
  263. void Free(void *ptr)
  264. {//=================
  265. if(ptr != NULL)
  266. free(ptr);
  267. }
  268. static void init_path(const char *path)
  269. {//====================================
  270. #ifdef PLATFORM_WINDOWS
  271. HKEY RegKey;
  272. unsigned long size;
  273. unsigned long var_type;
  274. char *env;
  275. unsigned char buf[sizeof(path_home)-13];
  276. if(path != NULL)
  277. {
  278. sprintf(path_home,"%s/espeak-data",path);
  279. return;
  280. }
  281. if((env = getenv("ESPEAK_DATA_PATH")) != NULL)
  282. {
  283. sprintf(path_home,"%s/espeak-data",env);
  284. if(GetFileLength(path_home) == -2)
  285. return; // an espeak-data directory exists
  286. }
  287. buf[0] = 0;
  288. RegOpenKeyExA(HKEY_LOCAL_MACHINE, "Software\\Microsoft\\Speech\\Voices\\Tokens\\eSpeak", 0, KEY_READ, &RegKey);
  289. size = sizeof(buf);
  290. var_type = REG_SZ;
  291. RegQueryValueExA(RegKey, "path", 0, &var_type, buf, &size);
  292. sprintf(path_home,"%s\\espeak-data",buf);
  293. #else
  294. char *env;
  295. if(path != NULL)
  296. {
  297. snprintf(path_home,sizeof(path_home),"%s/espeak-data",path);
  298. return;
  299. }
  300. // check for environment variable
  301. if((env = getenv("ESPEAK_DATA_PATH")) != NULL)
  302. {
  303. snprintf(path_home,sizeof(path_home),"%s/espeak-data",env);
  304. if(GetFileLength(path_home) == -2)
  305. return; // an espeak-data directory exists
  306. }
  307. snprintf(path_home,sizeof(path_home),"%s/espeak-data",getenv("HOME"));
  308. if(access(path_home,R_OK) != 0)
  309. {
  310. strcpy(path_home,PATH_ESPEAK_DATA);
  311. }
  312. #endif
  313. }
  314. static int initialise(int control)
  315. {//===============================
  316. int param;
  317. int result;
  318. int srate = 22050; // default sample rate 22050 Hz
  319. err = EE_OK;
  320. LoadConfig();
  321. if((result = LoadPhData(&srate)) != 1) // reads sample rate from espeak-data/phontab
  322. {
  323. if(result == -1)
  324. {
  325. fprintf(stderr,"Failed to load espeak-data\n");
  326. if((control & espeakINITIALIZE_DONT_EXIT) == 0)
  327. {
  328. exit(1);
  329. }
  330. }
  331. else
  332. fprintf(stderr,"Wrong version of espeak-data 0x%x (expects 0x%x) at %s\n",result,version_phdata,path_home);
  333. }
  334. WavegenInit(srate,0);
  335. memset(&current_voice_selected,0,sizeof(current_voice_selected));
  336. SetVoiceStack(NULL, "");
  337. SynthesizeInit();
  338. InitNamedata();
  339. for(param=0; param<N_SPEECH_PARAM; param++)
  340. param_stack[0].parameter[param] = param_defaults[param];
  341. return(0);
  342. }
  343. static espeak_ERROR Synthesize(unsigned int unique_identifier, const void *text, int flags)
  344. {//========================================================================================
  345. // Fill the buffer with output sound
  346. int length;
  347. int finished = 0;
  348. int count_buffers = 0;
  349. #ifdef USE_ASYNC
  350. uint32_t a_write_pos=0;
  351. #endif
  352. #ifdef DEBUG_ENABLED
  353. ENTER("Synthesize");
  354. if (text)
  355. {
  356. SHOW("Synthesize > uid=%d, flags=%d, >>>text=%s<<<\n", unique_identifier, flags, text);
  357. }
  358. #endif
  359. if((outbuf==NULL) || (event_list==NULL))
  360. return(EE_INTERNAL_ERROR); // espeak_Initialize() has not been called
  361. option_multibyte = flags & 7;
  362. option_ssml = flags & espeakSSML;
  363. option_phoneme_input = flags & espeakPHONEMES;
  364. option_endpause = flags & espeakENDPAUSE;
  365. count_samples = 0;
  366. #ifdef USE_ASYNC
  367. if(my_mode == AUDIO_OUTPUT_PLAYBACK)
  368. {
  369. a_write_pos = wave_get_write_position(my_audio);
  370. }
  371. #endif
  372. if(translator == NULL)
  373. {
  374. SetVoiceByName("default");
  375. }
  376. SpeakNextClause(NULL,text,0);
  377. if(my_mode == AUDIO_OUTPUT_SYNCH_PLAYBACK)
  378. {
  379. for(;;)
  380. {
  381. #ifdef PLATFORM_WINDOWS
  382. Sleep(300); // 0.3s
  383. #else
  384. #ifdef USE_NANOSLEEP
  385. struct timespec period;
  386. struct timespec remaining;
  387. period.tv_sec = 0;
  388. period.tv_nsec = 300000000; // 0.3 sec
  389. nanosleep(&period,&remaining);
  390. #else
  391. sleep(1);
  392. #endif
  393. #endif
  394. if(SynthOnTimer() != 0)
  395. break;
  396. }
  397. return(EE_OK);
  398. }
  399. for(;;)
  400. {
  401. #ifdef DEBUG_ENABLED
  402. SHOW("Synthesize > %s\n","for (next)");
  403. #endif
  404. out_ptr = outbuf;
  405. out_end = &outbuf[outbuf_size];
  406. event_list_ix = 0;
  407. WavegenFill(0);
  408. length = (out_ptr - outbuf)/2;
  409. count_samples += length;
  410. event_list[event_list_ix].type = espeakEVENT_LIST_TERMINATED; // indicates end of event list
  411. event_list[event_list_ix].unique_identifier = my_unique_identifier;
  412. event_list[event_list_ix].user_data = my_user_data;
  413. count_buffers++;
  414. if (my_mode==AUDIO_OUTPUT_PLAYBACK)
  415. {
  416. #ifdef USE_ASYNC
  417. finished = create_events((short *)outbuf, length, event_list, a_write_pos);
  418. if(finished < 0)
  419. return EE_INTERNAL_ERROR;
  420. length = 0; // the wave data are played once.
  421. #endif
  422. }
  423. else
  424. {
  425. finished = synth_callback((short *)outbuf, length, event_list);
  426. }
  427. if(finished)
  428. {
  429. SpeakNextClause(NULL,0,2); // stop
  430. break;
  431. }
  432. if(Generate(phoneme_list,&n_phoneme_list,1)==0)
  433. {
  434. if(WcmdqUsed() == 0)
  435. {
  436. // don't process the next clause until the previous clause has finished generating speech.
  437. // This ensures that <audio> tag (which causes end-of-clause) is at a sound buffer boundary
  438. event_list[0].type = espeakEVENT_LIST_TERMINATED;
  439. event_list[0].unique_identifier = my_unique_identifier;
  440. event_list[0].user_data = my_user_data;
  441. if(SpeakNextClause(NULL,NULL,1)==0)
  442. {
  443. #ifdef USE_ASYNC
  444. if (my_mode==AUDIO_OUTPUT_PLAYBACK)
  445. {
  446. if(dispatch_audio(NULL, 0, NULL) < 0) // TBD: test case
  447. return err = EE_INTERNAL_ERROR;
  448. }
  449. else
  450. {
  451. synth_callback(NULL, 0, event_list); // NULL buffer ptr indicates end of data
  452. }
  453. #else
  454. synth_callback(NULL, 0, event_list); // NULL buffer ptr indicates end of data
  455. #endif
  456. break;
  457. }
  458. }
  459. }
  460. }
  461. return(EE_OK);
  462. } // end of Synthesize
  463. #ifdef DEBUG_ENABLED
  464. static const char* label[] = {
  465. "END_OF_EVENT_LIST",
  466. "WORD",
  467. "SENTENCE",
  468. "MARK",
  469. "PLAY",
  470. "END",
  471. "MSG_TERMINATED",
  472. "PHONEME",
  473. "SAMPLERATE",
  474. "??" };
  475. #endif
  476. void MarkerEvent(int type, unsigned int char_position, int value, int value2, unsigned char *out_ptr)
  477. {//==================================================================================================
  478. // type: 1=word, 2=sentence, 3=named mark, 4=play audio, 5=end, 7=phoneme
  479. ENTER("MarkerEvent");
  480. espeak_EVENT *ep;
  481. double time;
  482. if((event_list == NULL) || (event_list_ix >= (n_event_list-2)))
  483. return;
  484. ep = &event_list[event_list_ix++];
  485. ep->type = (espeak_EVENT_TYPE)type;
  486. ep->unique_identifier = my_unique_identifier;
  487. ep->user_data = my_user_data;
  488. ep->text_position = char_position & 0xffffff;
  489. ep->length = char_position >> 24;
  490. time = (double(count_samples + mbrola_delay + (out_ptr - out_start)/2)*1000.0)/samplerate;
  491. ep->audio_position = int(time);
  492. ep->sample = (count_samples + mbrola_delay + (out_ptr - out_start)/2);
  493. #ifdef DEBUG_ENABLED
  494. SHOW("MarkerEvent > count_samples=%d, out_ptr=%x, out_start=0x%x\n",count_samples, out_ptr, out_start);
  495. SHOW("*** MarkerEvent > type=%s, uid=%d, text_pos=%d, length=%d, audio_position=%d, sample=%d\n",
  496. label[ep->type], ep->unique_identifier, ep->text_position, ep->length,
  497. ep->audio_position, ep->sample);
  498. #endif
  499. if((type == espeakEVENT_MARK) || (type == espeakEVENT_PLAY))
  500. ep->id.name = &namedata[value];
  501. else
  502. //#ifdef deleted
  503. // temporarily removed, don't introduce until after eSpeak version 1.46.02
  504. if(type == espeakEVENT_PHONEME)
  505. {
  506. int *p;
  507. p = (int *)(ep->id.string);
  508. p[0] = value;
  509. p[1] = value2;
  510. }
  511. else
  512. //#endif
  513. {
  514. ep->id.number = value;
  515. }
  516. } // end of MarkerEvent
  517. espeak_ERROR sync_espeak_Synth(unsigned int unique_identifier, const void *text, size_t size,
  518. unsigned int position, espeak_POSITION_TYPE position_type,
  519. unsigned int end_position, unsigned int flags, void* user_data)
  520. {//===========================================================================
  521. #ifdef DEBUG_ENABLED
  522. ENTER("sync_espeak_Synth");
  523. SHOW("sync_espeak_Synth > position=%d, position_type=%d, end_position=%d, flags=%d, user_data=0x%x, text=%s\n", position, position_type, end_position, flags, user_data, text);
  524. #endif
  525. espeak_ERROR aStatus;
  526. InitText(flags);
  527. my_unique_identifier = unique_identifier;
  528. my_user_data = user_data;
  529. for (int i=0; i < N_SPEECH_PARAM; i++)
  530. saved_parameters[i] = param_stack[0].parameter[i];
  531. switch(position_type)
  532. {
  533. case POS_CHARACTER:
  534. skip_characters = position;
  535. break;
  536. case POS_WORD:
  537. skip_words = position;
  538. break;
  539. case POS_SENTENCE:
  540. skip_sentences = position;
  541. break;
  542. }
  543. if(skip_characters || skip_words || skip_sentences)
  544. skipping_text = 1;
  545. end_character_position = end_position;
  546. aStatus = Synthesize(unique_identifier, text, flags);
  547. #ifdef USE_ASYNC
  548. wave_flush(my_audio);
  549. #endif
  550. SHOW_TIME("LEAVE sync_espeak_Synth");
  551. return aStatus;
  552. } // end of sync_espeak_Synth
  553. espeak_ERROR sync_espeak_Synth_Mark(unsigned int unique_identifier, const void *text, size_t size,
  554. const char *index_mark, unsigned int end_position,
  555. unsigned int flags, void* user_data)
  556. {//=========================================================================
  557. espeak_ERROR aStatus;
  558. InitText(flags);
  559. my_unique_identifier = unique_identifier;
  560. my_user_data = user_data;
  561. if(index_mark != NULL)
  562. {
  563. strncpy0(skip_marker, index_mark, sizeof(skip_marker));
  564. skipping_text = 1;
  565. }
  566. end_character_position = end_position;
  567. aStatus = Synthesize(unique_identifier, text, flags | espeakSSML);
  568. SHOW_TIME("LEAVE sync_espeak_Synth_Mark");
  569. return (aStatus);
  570. } // end of sync_espeak_Synth_Mark
  571. void sync_espeak_Key(const char *key)
  572. {//==================================
  573. // symbolic name, symbolicname_character - is there a system resource of symbolic names per language?
  574. int letter;
  575. int ix;
  576. ix = utf8_in(&letter,key);
  577. if(key[ix] == 0)
  578. {
  579. // a single character
  580. sync_espeak_Char(letter);
  581. return;
  582. }
  583. my_unique_identifier = 0;
  584. my_user_data = NULL;
  585. Synthesize(0, key,0); // speak key as a text string
  586. }
  587. void sync_espeak_Char(wchar_t character)
  588. {//=====================================
  589. // is there a system resource of character names per language?
  590. char buf[80];
  591. my_unique_identifier = 0;
  592. my_user_data = NULL;
  593. sprintf(buf,"<say-as interpret-as=\"tts:char\">&#%d;</say-as>",character);
  594. Synthesize(0, buf,espeakSSML);
  595. }
  596. void sync_espeak_SetPunctuationList(const wchar_t *punctlist)
  597. {//==========================================================
  598. // Set the list of punctuation which are spoken for "some".
  599. my_unique_identifier = 0;
  600. my_user_data = NULL;
  601. option_punctlist[0] = 0;
  602. if(punctlist != NULL)
  603. {
  604. wcsncpy(option_punctlist, punctlist, N_PUNCTLIST);
  605. option_punctlist[N_PUNCTLIST-1] = 0;
  606. }
  607. } // end of sync_espeak_SetPunctuationList
  608. #pragma GCC visibility push(default)
  609. ESPEAK_API void espeak_SetSynthCallback(t_espeak_callback* SynthCallback)
  610. {//======================================================================
  611. ENTER("espeak_SetSynthCallback");
  612. synth_callback = SynthCallback;
  613. #ifdef USE_ASYNC
  614. event_set_callback(synth_callback);
  615. #endif
  616. }
  617. ESPEAK_API void espeak_SetUriCallback(int (* UriCallback)(int, const char*, const char *))
  618. {//=======================================================================================
  619. ENTER("espeak_SetUriCallback");
  620. uri_callback = UriCallback;
  621. }
  622. ESPEAK_API void espeak_SetPhonemeCallback(int (* PhonemeCallback)(const char*))
  623. {//===========================================================================
  624. phoneme_callback = PhonemeCallback;
  625. }
  626. ESPEAK_API int espeak_Initialize(espeak_AUDIO_OUTPUT output_type, int buf_length, const char *path, int options)
  627. {//=============================================================================================================
  628. ENTER("espeak_Initialize");
  629. int param;
  630. // It seems that the wctype functions don't work until the locale has been set
  631. // to something other than the default "C". Then, not only Latin1 but also the
  632. // other characters give the correct results with iswalpha() etc.
  633. #ifdef PLATFORM_RISCOS
  634. setlocale(LC_CTYPE,"ISO8859-1");
  635. #else
  636. if(setlocale(LC_CTYPE,"C.UTF-8") == NULL)
  637. {
  638. if(setlocale(LC_CTYPE,"UTF-8") == NULL)
  639. if(setlocale(LC_CTYPE,"en_US.UTF-8") == NULL)
  640. setlocale(LC_CTYPE,"");
  641. }
  642. #endif
  643. init_path(path);
  644. initialise(options);
  645. select_output(output_type);
  646. if(f_logespeak)
  647. {
  648. fprintf(f_logespeak,"INIT mode %d options 0x%x\n",output_type,options);
  649. }
  650. // buflength is in mS, allocate 2 bytes per sample
  651. if((buf_length == 0) || (output_type == AUDIO_OUTPUT_PLAYBACK) || (output_type == AUDIO_OUTPUT_SYNCH_PLAYBACK))
  652. buf_length = 200;
  653. outbuf_size = (buf_length * samplerate)/500;
  654. outbuf = (unsigned char*)realloc(outbuf,outbuf_size);
  655. if((out_start = outbuf) == NULL)
  656. return(EE_INTERNAL_ERROR);
  657. // allocate space for event list. Allow 200 events per second.
  658. // Add a constant to allow for very small buf_length
  659. n_event_list = (buf_length*200)/1000 + 20;
  660. if((event_list = (espeak_EVENT *)realloc(event_list,sizeof(espeak_EVENT) * n_event_list)) == NULL)
  661. return(EE_INTERNAL_ERROR);
  662. option_phonemes = 0;
  663. option_mbrola_phonemes = 0;
  664. option_phoneme_events = (options & (espeakINITIALIZE_PHONEME_EVENTS | espeakINITIALIZE_PHONEME_IPA));
  665. VoiceReset(0);
  666. // SetVoiceByName("default");
  667. for(param=0; param<N_SPEECH_PARAM; param++)
  668. param_stack[0].parameter[param] = saved_parameters[param] = param_defaults[param];
  669. SetParameter(espeakRATE,175,0);
  670. SetParameter(espeakVOLUME,100,0);
  671. SetParameter(espeakCAPITALS,option_capitals,0);
  672. SetParameter(espeakPUNCTUATION,option_punctuation,0);
  673. SetParameter(espeakWORDGAP,0,0);
  674. // DoVoiceChange(voice);
  675. #ifdef USE_ASYNC
  676. fifo_init();
  677. #endif
  678. return(samplerate);
  679. }
  680. ESPEAK_API espeak_ERROR espeak_Synth(const void *text, size_t size,
  681. unsigned int position,
  682. espeak_POSITION_TYPE position_type,
  683. unsigned int end_position, unsigned int flags,
  684. unsigned int* unique_identifier, void* user_data)
  685. {//=====================================================================================
  686. #ifdef DEBUG_ENABLED
  687. ENTER("espeak_Synth");
  688. SHOW("espeak_Synth > position=%d, position_type=%d, end_position=%d, flags=%d, user_data=0x%x, text=%s\n", position, position_type, end_position, flags, user_data, text);
  689. #endif
  690. if(f_logespeak)
  691. {
  692. fprintf(f_logespeak,"\nSYNTH posn %d %d %d flags 0x%x\n%s\n",position,end_position,position_type,flags, (const char *)text);
  693. fflush(f_logespeak);
  694. }
  695. espeak_ERROR a_error=EE_INTERNAL_ERROR;
  696. static unsigned int temp_identifier;
  697. if (unique_identifier == NULL)
  698. {
  699. unique_identifier = &temp_identifier;
  700. }
  701. *unique_identifier = 0;
  702. if(synchronous_mode)
  703. {
  704. return(sync_espeak_Synth(0,text,size,position,position_type,end_position,flags,user_data));
  705. }
  706. #ifdef USE_ASYNC
  707. // Create the text command
  708. t_espeak_command* c1 = create_espeak_text(text, size, position, position_type, end_position, flags, user_data);
  709. // Retrieve the unique identifier
  710. *unique_identifier = c1->u.my_text.unique_identifier;
  711. // Create the "terminated msg" command (same uid)
  712. t_espeak_command* c2 = create_espeak_terminated_msg(*unique_identifier, user_data);
  713. // Try to add these 2 commands (single transaction)
  714. if (c1 && c2)
  715. {
  716. a_error = fifo_add_commands(c1, c2);
  717. if (a_error != EE_OK)
  718. {
  719. delete_espeak_command(c1);
  720. delete_espeak_command(c2);
  721. c1=c2=NULL;
  722. }
  723. }
  724. else
  725. {
  726. delete_espeak_command(c1);
  727. delete_espeak_command(c2);
  728. }
  729. #endif
  730. return a_error;
  731. } // end of espeak_Synth
  732. ESPEAK_API espeak_ERROR espeak_Synth_Mark(const void *text, size_t size,
  733. const char *index_mark,
  734. unsigned int end_position,
  735. unsigned int flags,
  736. unsigned int* unique_identifier,
  737. void* user_data)
  738. {//=========================================================================
  739. #ifdef DEBUG_ENABLED
  740. ENTER("espeak_Synth_Mark");
  741. SHOW("espeak_Synth_Mark > index_mark=%s, end_position=%d, flags=%d, text=%s\n", index_mark, end_position, flags, text);
  742. #endif
  743. espeak_ERROR a_error=EE_OK;
  744. static unsigned int temp_identifier;
  745. if(f_logespeak)
  746. {
  747. fprintf(f_logespeak,"\nSYNTH MARK %s posn %d flags 0x%x\n%s\n",index_mark,end_position,flags, (const char *)text);
  748. }
  749. if (unique_identifier == NULL)
  750. {
  751. unique_identifier = &temp_identifier;
  752. }
  753. *unique_identifier = 0;
  754. if(synchronous_mode)
  755. {
  756. return(sync_espeak_Synth_Mark(0,text,size,index_mark,end_position,flags,user_data));
  757. }
  758. #ifdef USE_ASYNC
  759. // Create the mark command
  760. t_espeak_command* c1 = create_espeak_mark(text, size, index_mark, end_position,
  761. flags, user_data);
  762. // Retrieve the unique identifier
  763. *unique_identifier = c1->u.my_mark.unique_identifier;
  764. // Create the "terminated msg" command (same uid)
  765. t_espeak_command* c2 = create_espeak_terminated_msg(*unique_identifier, user_data);
  766. // Try to add these 2 commands (single transaction)
  767. if (c1 && c2)
  768. {
  769. a_error = fifo_add_commands(c1, c2);
  770. if (a_error != EE_OK)
  771. {
  772. delete_espeak_command(c1);
  773. delete_espeak_command(c2);
  774. c1=c2=NULL;
  775. }
  776. }
  777. else
  778. {
  779. delete_espeak_command(c1);
  780. delete_espeak_command(c2);
  781. }
  782. #endif
  783. return a_error;
  784. } // end of espeak_Synth_Mark
  785. ESPEAK_API espeak_ERROR espeak_Key(const char *key)
  786. {//================================================
  787. ENTER("espeak_Key");
  788. // symbolic name, symbolicname_character - is there a system resource of symbolicnames per language
  789. if(f_logespeak)
  790. {
  791. fprintf(f_logespeak,"\nKEY %s\n",key);
  792. }
  793. espeak_ERROR a_error = EE_OK;
  794. if(synchronous_mode)
  795. {
  796. sync_espeak_Key(key);
  797. return(EE_OK);
  798. }
  799. #ifdef USE_ASYNC
  800. t_espeak_command* c = create_espeak_key( key, NULL);
  801. a_error = fifo_add_command(c);
  802. if (a_error != EE_OK)
  803. {
  804. delete_espeak_command(c);
  805. }
  806. #endif
  807. return a_error;
  808. }
  809. ESPEAK_API espeak_ERROR espeak_Char(wchar_t character)
  810. {//===========================================
  811. ENTER("espeak_Char");
  812. // is there a system resource of character names per language?
  813. if(f_logespeak)
  814. {
  815. fprintf(f_logespeak,"\nCHAR U+%x\n",character);
  816. }
  817. #ifdef USE_ASYNC
  818. espeak_ERROR a_error;
  819. if(synchronous_mode)
  820. {
  821. sync_espeak_Char(character);
  822. return(EE_OK);
  823. }
  824. t_espeak_command* c = create_espeak_char( character, NULL);
  825. a_error = fifo_add_command(c);
  826. if (a_error != EE_OK)
  827. {
  828. delete_espeak_command(c);
  829. }
  830. return a_error;
  831. #else
  832. sync_espeak_Char(character);
  833. return(EE_OK);
  834. #endif
  835. }
  836. ESPEAK_API espeak_ERROR espeak_SetVoiceByName(const char *name)
  837. {//============================================================
  838. ENTER("espeak_SetVoiceByName");
  839. //#ifdef USE_ASYNC
  840. // I don't think there's a need to queue change voice requests
  841. #ifdef deleted
  842. espeak_ERROR a_error;
  843. if(synchronous_mode)
  844. {
  845. return(SetVoiceByName(name));
  846. }
  847. t_espeak_command* c = create_espeak_voice_name(name);
  848. a_error = fifo_add_command(c);
  849. if (a_error != EE_OK)
  850. {
  851. delete_espeak_command(c);
  852. }
  853. return a_error;
  854. #else
  855. return(SetVoiceByName(name));
  856. #endif
  857. } // end of espeak_SetVoiceByName
  858. ESPEAK_API espeak_ERROR espeak_SetVoiceByProperties(espeak_VOICE *voice_selector)
  859. {//==============================================================================
  860. ENTER("espeak_SetVoiceByProperties");
  861. //#ifdef USE_ASYNC
  862. #ifdef deleted
  863. espeak_ERROR a_error;
  864. if(synchronous_mode)
  865. {
  866. return(SetVoiceByProperties(voice_selector));
  867. }
  868. t_espeak_command* c = create_espeak_voice_spec( voice_selector);
  869. a_error = fifo_add_command(c);
  870. if (a_error != EE_OK)
  871. {
  872. delete_espeak_command(c);
  873. }
  874. return a_error;
  875. #else
  876. return(SetVoiceByProperties(voice_selector));
  877. #endif
  878. } // end of espeak_SetVoiceByProperties
  879. ESPEAK_API int espeak_GetParameter(espeak_PARAMETER parameter, int current)
  880. {//========================================================================
  881. ENTER("espeak_GetParameter");
  882. // current: 0=default value, 1=current value
  883. if(current)
  884. {
  885. return(param_stack[0].parameter[parameter]);
  886. }
  887. else
  888. {
  889. return(param_defaults[parameter]);
  890. }
  891. } // end of espeak_GetParameter
  892. ESPEAK_API espeak_ERROR espeak_SetParameter(espeak_PARAMETER parameter, int value, int relative)
  893. {//=============================================================================================
  894. ENTER("espeak_SetParameter");
  895. if(f_logespeak)
  896. {
  897. fprintf(f_logespeak,"SETPARAM %d %d %d\n",parameter,value,relative);
  898. }
  899. #ifdef USE_ASYNC
  900. espeak_ERROR a_error;
  901. if(synchronous_mode)
  902. {
  903. SetParameter(parameter,value,relative);
  904. return(EE_OK);
  905. }
  906. t_espeak_command* c = create_espeak_parameter(parameter, value, relative);
  907. a_error = fifo_add_command(c);
  908. if (a_error != EE_OK)
  909. {
  910. delete_espeak_command(c);
  911. }
  912. return a_error;
  913. #else
  914. SetParameter(parameter,value,relative);
  915. return(EE_OK);
  916. #endif
  917. }
  918. ESPEAK_API espeak_ERROR espeak_SetPunctuationList(const wchar_t *punctlist)
  919. {//================================================================
  920. ENTER("espeak_SetPunctuationList");
  921. // Set the list of punctuation which are spoken for "some".
  922. #ifdef USE_ASYNC
  923. espeak_ERROR a_error;
  924. if(synchronous_mode)
  925. {
  926. sync_espeak_SetPunctuationList(punctlist);
  927. return(EE_OK);
  928. }
  929. t_espeak_command* c = create_espeak_punctuation_list( punctlist);
  930. a_error = fifo_add_command(c);
  931. if (a_error != EE_OK)
  932. {
  933. delete_espeak_command(c);
  934. }
  935. return a_error;
  936. #else
  937. sync_espeak_SetPunctuationList(punctlist);
  938. return(EE_OK);
  939. #endif
  940. } // end of espeak_SetPunctuationList
  941. ESPEAK_API void espeak_SetPhonemeTrace(int value, FILE *stream)
  942. {//============================================================
  943. ENTER("espeak_SetPhonemes");
  944. /* Controls the output of phoneme symbols for the text
  945. bits 0-3:
  946. value=0 No phoneme output (default)
  947. value=1 Output the translated phoneme symbols for the text
  948. value=2 as (1), but also output a trace of how the translation was done (matching rules and list entries)
  949. value=3 as (1), but produces IPA phoneme names rather than ascii
  950. bit 4: produce mbrola pho data
  951. */
  952. option_phonemes = value & 7;
  953. option_mbrola_phonemes = value & 16;
  954. f_trans = stream;
  955. if(stream == NULL)
  956. f_trans = stderr;
  957. } // end of espeak_SetPhonemes
  958. ESPEAK_API const char *espeak_TextToPhonemes(const void **textptr, int textmode, int phonememode)
  959. {//=================================================================================================
  960. /* phoneme_mode bits 0-3: 0=only phoneme names, 1=ties, 2=ZWJ, 3=underscore separator
  961. bits 4-7: 0=eSpeak phoneme names, 1=IPA
  962. */
  963. option_multibyte = textmode & 7;
  964. *textptr = TranslateClause(translator, NULL, *textptr, NULL, NULL);
  965. return(GetTranslatedPhonemeString(phonememode));
  966. }
  967. ESPEAK_API void espeak_CompileDictionary(const char *path, FILE *log, int flags)
  968. {//=============================================================================
  969. ENTER("espeak_CompileDictionary");
  970. CompileDictionary(path, dictionary_name, log, NULL, flags);
  971. } // end of espeak_CompileDirectory
  972. ESPEAK_API espeak_ERROR espeak_Cancel(void)
  973. {//===============================
  974. #ifdef USE_ASYNC
  975. ENTER("espeak_Cancel");
  976. fifo_stop();
  977. event_clear_all();
  978. if(my_mode == AUDIO_OUTPUT_PLAYBACK)
  979. {
  980. wave_close(my_audio);
  981. }
  982. SHOW_TIME("espeak_Cancel > LEAVE");
  983. #endif
  984. embedded_value[EMBED_T] = 0; // reset echo for pronunciation announcements
  985. for (int i=0; i < N_SPEECH_PARAM; i++)
  986. SetParameter(i, saved_parameters[i], 0);
  987. return EE_OK;
  988. } // end of espeak_Cancel
  989. ESPEAK_API int espeak_IsPlaying(void)
  990. {//==================================
  991. // ENTER("espeak_IsPlaying");
  992. #ifdef USE_ASYNC
  993. if((my_mode == AUDIO_OUTPUT_PLAYBACK) && wave_is_busy(my_audio))
  994. return(1);
  995. return(fifo_is_busy());
  996. #else
  997. return(0);
  998. #endif
  999. } // end of espeak_IsPlaying
  1000. ESPEAK_API espeak_ERROR espeak_Synchronize(void)
  1001. {//=============================================
  1002. espeak_ERROR berr = err;
  1003. #ifdef USE_ASYNC
  1004. SHOW_TIME("espeak_Synchronize > ENTER");
  1005. while (espeak_IsPlaying())
  1006. {
  1007. usleep(20000);
  1008. }
  1009. #endif
  1010. err = EE_OK;
  1011. SHOW_TIME("espeak_Synchronize > LEAVE");
  1012. return berr;
  1013. } // end of espeak_Synchronize
  1014. extern void FreePhData(void);
  1015. extern void FreeVoiceList(void);
  1016. ESPEAK_API espeak_ERROR espeak_Terminate(void)
  1017. {//===========================================
  1018. ENTER("espeak_Terminate");
  1019. #ifdef USE_ASYNC
  1020. fifo_stop();
  1021. fifo_terminate();
  1022. event_terminate();
  1023. if(my_mode == AUDIO_OUTPUT_PLAYBACK)
  1024. {
  1025. wave_close(my_audio);
  1026. wave_terminate();
  1027. out_samplerate = 0;
  1028. }
  1029. #endif
  1030. Free(event_list);
  1031. event_list = NULL;
  1032. Free(outbuf);
  1033. outbuf = NULL;
  1034. FreePhData();
  1035. FreeVoiceList();
  1036. if(f_logespeak)
  1037. {
  1038. fclose(f_logespeak);
  1039. f_logespeak = NULL;
  1040. }
  1041. return EE_OK;
  1042. } // end of espeak_Terminate
  1043. ESPEAK_API const char *espeak_Info(const char **ptr)
  1044. {//=================================================
  1045. if(ptr != NULL)
  1046. {
  1047. *ptr = path_home;
  1048. }
  1049. return(version_string);
  1050. }
  1051. #pragma GCC visibility pop