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.

eSpeakService.cpp 12KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409
  1. /*
  2. * Copyright (C) 2012-2013 Reece H. Dunn
  3. * Copyright (C) 2011 Google Inc.
  4. *
  5. * Licensed under the Apache License, Version 2.0 (the "License");
  6. * you may not use this file except in compliance with the License.
  7. * You may obtain a copy of the License at
  8. *
  9. * http://www.apache.org/licenses/LICENSE-2.0
  10. *
  11. * Unless required by applicable law or agreed to in writing, software
  12. * distributed under the License is distributed on an "AS IS" BASIS,
  13. * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  14. * See the License for the specific language governing permissions and
  15. * limitations under the License.
  16. */
  17. /*
  18. * This file contains the JNI bindings to eSpeak used by SpeechSynthesis.java.
  19. *
  20. * Android Version: 4.0 (Ice Cream Sandwich)
  21. * API Version: 14
  22. */
  23. #include <stdio.h>
  24. #include <unistd.h>
  25. #include <stdlib.h>
  26. #include <jni.h>
  27. #include <speak_lib.h>
  28. #include <Log.h>
  29. /** @name Java to Wide String Helpers
  30. * @brief These are helpers for converting a jstring to wchar_t*.
  31. *
  32. * This assumes that wchar_t is a 32-bit (UTF-32) value.
  33. */
  34. //@{
  35. const char *utf8_read(const char *in, wchar_t &c)
  36. {
  37. if (uint8_t(*in) < 0x80)
  38. c = *in++;
  39. else switch (uint8_t(*in) & 0xF0)
  40. {
  41. default:
  42. c = uint8_t(*in++) & 0x1F;
  43. c = (c << 6) + (uint8_t(*in++) & 0x3F);
  44. break;
  45. case 0xE0:
  46. c = uint8_t(*in++) & 0x0F;
  47. c = (c << 6) + (uint8_t(*in++) & 0x3F);
  48. c = (c << 6) + (uint8_t(*in++) & 0x3F);
  49. break;
  50. case 0xF0:
  51. c = uint8_t(*in++) & 0x07;
  52. c = (c << 6) + (uint8_t(*in++) & 0x3F);
  53. c = (c << 6) + (uint8_t(*in++) & 0x3F);
  54. c = (c << 6) + (uint8_t(*in++) & 0x3F);
  55. break;
  56. }
  57. return in;
  58. }
  59. class unicode_string
  60. {
  61. static_assert(sizeof(wchar_t) == 4, "wchar_t is not UTF-32");
  62. public:
  63. unicode_string(JNIEnv *env, jstring str);
  64. ~unicode_string();
  65. const wchar_t *c_str() const { return mString; }
  66. private:
  67. wchar_t *mString;
  68. };
  69. unicode_string::unicode_string(JNIEnv *env, jstring str)
  70. : mString(NULL)
  71. {
  72. if (str == NULL) return;
  73. const char *utf8 = env->GetStringUTFChars(str, NULL);
  74. mString = (wchar_t *)malloc((strlen(utf8) + 1) * sizeof(wchar_t));
  75. const char *utf8_current = utf8;
  76. wchar_t *utf32_current = mString;
  77. while (*utf8_current)
  78. {
  79. utf8_current = utf8_read(utf8_current, *utf32_current);
  80. ++utf32_current;
  81. }
  82. *utf32_current = 0;
  83. env->ReleaseStringUTFChars(str, utf8);
  84. }
  85. unicode_string::~unicode_string()
  86. {
  87. if (mString) free(mString);
  88. }
  89. //@}
  90. #define LOG_TAG "eSpeakService"
  91. #define DEBUG true
  92. enum synthesis_result {
  93. SYNTH_CONTINUE = 0,
  94. SYNTH_ABORT = 1
  95. };
  96. const int DEFAULT_BUFFER_SIZE = 1000;
  97. struct native_data_t {
  98. JNIEnv *env;
  99. jobject object;
  100. int sampleRate;
  101. int bufferSizeInMillis;
  102. native_data_t() {
  103. env = NULL;
  104. object = NULL;
  105. sampleRate = 0;
  106. bufferSizeInMillis = DEFAULT_BUFFER_SIZE;
  107. }
  108. };
  109. jmethodID METHOD_nativeSynthCallback;
  110. jfieldID FIELD_mNativeData;
  111. static inline native_data_t *getNativeData(JNIEnv *env, jobject object) {
  112. return (native_data_t *) (env->GetIntField(object, FIELD_mNativeData));
  113. }
  114. /* Callback from espeak. Should call back to the TTS API */
  115. static int SynthCallback(short *audioData, int numSamples,
  116. espeak_EVENT *events) {
  117. native_data_t *nat = (native_data_t *) events->user_data;
  118. JNIEnv *env = nat->env;
  119. jobject object = nat->object;
  120. if (numSamples < 1) {
  121. env->CallVoidMethod(object, METHOD_nativeSynthCallback, NULL);
  122. return SYNTH_ABORT;
  123. } else {
  124. jbyteArray arrayAudioData = env->NewByteArray(numSamples * 2);
  125. env->SetByteArrayRegion(arrayAudioData, 0, (numSamples * 2), (jbyte *) audioData);
  126. env->CallVoidMethod(object, METHOD_nativeSynthCallback, arrayAudioData);
  127. return SYNTH_CONTINUE;
  128. }
  129. }
  130. #ifdef __cplusplus
  131. extern "C" {
  132. #endif /* __cplusplus */
  133. JNIEXPORT jint
  134. JNICALL JNI_OnLoad(JavaVM *vm, void *reserved) {
  135. JNIEnv *env;
  136. if (vm->GetEnv((void **) &env, JNI_VERSION_1_6) != JNI_OK) {
  137. LOGE("Failed to get the environment using GetEnv()");
  138. return -1;
  139. }
  140. return JNI_VERSION_1_6;
  141. }
  142. JNIEXPORT jboolean
  143. JNICALL Java_com_reecedunn_espeak_SpeechSynthesis_nativeClassInit(
  144. JNIEnv* env, jclass clazz) {
  145. if (DEBUG) LOGV("%s", __FUNCTION__);
  146. METHOD_nativeSynthCallback = env->GetMethodID(clazz, "nativeSynthCallback", "([B)V");
  147. FIELD_mNativeData = env->GetFieldID(clazz, "mNativeData", "I");
  148. return JNI_TRUE;
  149. }
  150. JNIEXPORT jboolean
  151. JNICALL Java_com_reecedunn_espeak_SpeechSynthesis_nativeCreate(
  152. JNIEnv *env, jobject object, jstring path) {
  153. if (DEBUG) LOGV("%s [env=%p, object=%p]", __FUNCTION__, env, object);
  154. native_data_t *nat = new native_data_t;
  155. if (nat == NULL) {
  156. LOGE("%s: out of memory!", __FUNCTION__);
  157. return JNI_FALSE;
  158. }
  159. env->SetIntField(object, FIELD_mNativeData, (jint) nat);
  160. const char *c_path = path ? env->GetStringUTFChars(path, NULL) : NULL;
  161. nat->object = env->NewWeakGlobalRef(object);
  162. if (DEBUG) LOGV("Initializing with path %s", c_path);
  163. nat->sampleRate = espeak_Initialize(AUDIO_OUTPUT_SYNCHRONOUS, nat->bufferSizeInMillis, c_path, 0);
  164. if (c_path) env->ReleaseStringUTFChars(path, c_path);
  165. return (nat->sampleRate > 0) ? JNI_TRUE : JNI_FALSE;
  166. }
  167. JNIEXPORT jboolean
  168. JNICALL Java_com_reecedunn_espeak_SpeechSynthesis_nativeDestroy(
  169. JNIEnv *env, jobject object) {
  170. if (DEBUG) LOGV("%s [env=%p, object=%p]", __FUNCTION__, env, object);
  171. native_data_t *nat = getNativeData(env, object);
  172. if (nat) {
  173. env->DeleteWeakGlobalRef(nat->object);
  174. delete nat;
  175. }
  176. return JNI_TRUE;
  177. }
  178. JNIEXPORT jobject
  179. JNICALL Java_com_reecedunn_espeak_SpeechSynthesis_nativeGetVersion(
  180. JNIEnv *env, jclass clazz) {
  181. if (DEBUG) LOGV("%s", __FUNCTION__);
  182. return env->NewStringUTF(espeak_Info(NULL));
  183. }
  184. JNIEXPORT jint
  185. JNICALL Java_com_reecedunn_espeak_SpeechSynthesis_nativeGetSampleRate(
  186. JNIEnv *env, jobject object) {
  187. if (DEBUG) LOGV("%s", __FUNCTION__);
  188. const native_data_t *nat = getNativeData(env, object);
  189. return (jint)(nat ? nat->sampleRate : 0);
  190. }
  191. JNIEXPORT jint
  192. JNICALL Java_com_reecedunn_espeak_SpeechSynthesis_nativeGetBufferSizeInMillis(
  193. JNIEnv *env, jobject object) {
  194. if (DEBUG) LOGV("%s", __FUNCTION__);
  195. const native_data_t *nat = getNativeData(env, object);
  196. return (jint) nat->bufferSizeInMillis;
  197. }
  198. JNIEXPORT jobjectArray
  199. JNICALL Java_com_reecedunn_espeak_SpeechSynthesis_nativeGetAvailableVoices(
  200. JNIEnv *env, jobject object) {
  201. if (DEBUG) LOGV("%s", __FUNCTION__);
  202. const espeak_VOICE **voices = espeak_ListVoices(NULL);
  203. int count;
  204. // First, count the number of voices returned.
  205. for (count = 0; voices[count] != NULL; count++);
  206. // Next, create a Java String array.
  207. jobjectArray voicesArray = (jobjectArray) env->NewObjectArray(
  208. count * 4, env->FindClass("java/lang/String"), NULL);
  209. const espeak_VOICE *v;
  210. char gender_buf[12];
  211. char age_buf[12];
  212. // Finally, populate the array.
  213. for (int i = 0, voicesIndex = 0; (v = voices[i]) != NULL; i++) {
  214. const char *lang_name = v->languages + 1;
  215. const char *identifier = v->identifier;
  216. sprintf(gender_buf, "%d", v->gender);
  217. sprintf(age_buf, "%d", v->age);
  218. env->SetObjectArrayElement(
  219. voicesArray, voicesIndex++, env->NewStringUTF(lang_name));
  220. env->SetObjectArrayElement(
  221. voicesArray, voicesIndex++, env->NewStringUTF(identifier));
  222. env->SetObjectArrayElement(
  223. voicesArray, voicesIndex++, env->NewStringUTF(gender_buf));
  224. env->SetObjectArrayElement(
  225. voicesArray, voicesIndex++, env->NewStringUTF(age_buf));
  226. }
  227. return voicesArray;
  228. }
  229. JNIEXPORT jboolean
  230. JNICALL Java_com_reecedunn_espeak_SpeechSynthesis_nativeSetVoiceByName(
  231. JNIEnv *env, jobject object, jstring name) {
  232. const char *c_name = name ? env->GetStringUTFChars(name, NULL) : NULL;
  233. if (DEBUG) LOGV("%s(name=%s)", __FUNCTION__, c_name);
  234. const espeak_ERROR result = espeak_SetVoiceByName(c_name);
  235. if (c_name) env->ReleaseStringUTFChars(name, c_name);
  236. switch (result) {
  237. case EE_OK: return JNI_TRUE;
  238. case EE_INTERNAL_ERROR: LOGE("espeak_SetVoiceByName: internal error."); break;
  239. case EE_BUFFER_FULL: LOGE("espeak_SetVoiceByName: buffer full."); break;
  240. case EE_NOT_FOUND: LOGE("espeak_SetVoiceByName: not found."); break;
  241. }
  242. return JNI_FALSE;
  243. }
  244. JNIEXPORT jboolean
  245. JNICALL Java_com_reecedunn_espeak_SpeechSynthesis_nativeSetVoiceByProperties(
  246. JNIEnv *env, jobject object, jstring language, jint gender, jint age) {
  247. const char *c_language = language ? env->GetStringUTFChars(language, NULL) : NULL;
  248. if (DEBUG) LOGV("%s(language=%s, gender=%d, age=%d)", __FUNCTION__, c_language, gender, age);
  249. espeak_VOICE voice_select;
  250. memset(&voice_select, 0, sizeof(espeak_VOICE));
  251. voice_select.languages = c_language;
  252. voice_select.gender = (int) gender;
  253. voice_select.age = (int) age;
  254. const espeak_ERROR result = espeak_SetVoiceByProperties(&voice_select);
  255. if (c_language) env->ReleaseStringUTFChars(language, c_language);
  256. switch (result) {
  257. case EE_OK: return JNI_TRUE;
  258. case EE_INTERNAL_ERROR: LOGE("espeak_SetVoiceByProperties: internal error."); break;
  259. case EE_BUFFER_FULL: LOGE("espeak_SetVoiceByProperties: buffer full."); break;
  260. case EE_NOT_FOUND: LOGE("espeak_SetVoiceByProperties: not found."); break;
  261. }
  262. return JNI_FALSE;
  263. }
  264. JNIEXPORT jboolean
  265. JNICALL Java_com_reecedunn_espeak_SpeechSynthesis_nativeSetParameter(
  266. JNIEnv *env, jobject object, jint parameter, jint value) {
  267. if (DEBUG) LOGV("%s(parameter=%d, value=%d)", __FUNCTION__, parameter, value);
  268. const espeak_ERROR result = espeak_SetParameter((espeak_PARAMETER)parameter, (int)value, 0);
  269. switch (result) {
  270. case EE_OK: return JNI_TRUE;
  271. case EE_INTERNAL_ERROR: LOGE("espeak_SetParameter: internal error."); break;
  272. case EE_BUFFER_FULL: LOGE("espeak_SetParameter: buffer full."); break;
  273. case EE_NOT_FOUND: LOGE("espeak_SetParameter: not found."); break;
  274. }
  275. return JNI_FALSE;
  276. }
  277. JNIEXPORT jint
  278. JNICALL Java_com_reecedunn_espeak_SpeechSynthesis_nativeGetParameter(
  279. JNIEnv *env, jobject object, jint parameter, jint current) {
  280. if (DEBUG) LOGV("%s(parameter=%d, pitch=%d)", __FUNCTION__, parameter, current);
  281. return espeak_GetParameter((espeak_PARAMETER)parameter, (int)current);
  282. }
  283. JNIEXPORT jboolean
  284. JNICALL Java_com_reecedunn_espeak_SpeechSynthesis_nativeSetPunctuationCharacters(
  285. JNIEnv *env, jobject object, jstring characters) {
  286. if (DEBUG) LOGV("%s)", __FUNCTION__);
  287. unicode_string list(env, characters);
  288. const espeak_ERROR result = espeak_SetPunctuationList(list.c_str());
  289. switch (result) {
  290. case EE_OK: return JNI_TRUE;
  291. case EE_INTERNAL_ERROR: LOGE("espeak_SetPunctuationList: internal error."); break;
  292. case EE_BUFFER_FULL: LOGE("espeak_SetPunctuationList: buffer full."); break;
  293. case EE_NOT_FOUND: LOGE("espeak_SetPunctuationList: not found."); break;
  294. }
  295. return JNI_FALSE;
  296. }
  297. JNIEXPORT jboolean
  298. JNICALL Java_com_reecedunn_espeak_SpeechSynthesis_nativeSynthesize(
  299. JNIEnv *env, jobject object, jstring text, jboolean isSsml) {
  300. if (DEBUG) LOGV("%s", __FUNCTION__);
  301. native_data_t *nat = getNativeData(env, object);
  302. const char *c_text = text ? env->GetStringUTFChars(text, NULL) : NULL;
  303. unsigned int unique_identifier;
  304. nat->env = env;
  305. espeak_SetSynthCallback(SynthCallback);
  306. const espeak_ERROR result = espeak_Synth(c_text, strlen(c_text), 0, // position
  307. POS_CHARACTER, 0, // end position (0 means no end position)
  308. isSsml ? espeakCHARS_UTF8 | espeakSSML // UTF-8 encoded SSML
  309. : espeakCHARS_UTF8, // UTF-8 encoded text
  310. &unique_identifier, nat);
  311. espeak_Synchronize();
  312. if (c_text) env->ReleaseStringUTFChars(text, c_text);
  313. switch (result) {
  314. case EE_OK: return JNI_TRUE;
  315. case EE_INTERNAL_ERROR: LOGE("espeak_Synth: internal error."); break;
  316. case EE_BUFFER_FULL: LOGE("espeak_Synth: buffer full."); break;
  317. case EE_NOT_FOUND: LOGE("espeak_Synth: not found."); break;
  318. }
  319. return JNI_TRUE;
  320. }
  321. JNIEXPORT jboolean
  322. JNICALL Java_com_reecedunn_espeak_SpeechSynthesis_nativeStop(
  323. JNIEnv *env, jobject object) {
  324. if (DEBUG) LOGV("%s", __FUNCTION__);
  325. espeak_Cancel();
  326. return JNI_TRUE;
  327. }
  328. #ifdef __cplusplus
  329. }
  330. #endif /* __cplusplus */