| 1 | // Copyright (C) 2015 The Qt Company Ltd. |
| 2 | // SPDX-License-Identifier: LicenseRef-Qt-Commercial OR LGPL-3.0-only OR GPL-2.0-only OR GPL-3.0-only |
| 3 | |
| 4 | |
| 5 | #include "qtexttospeech_speechd.h" |
| 6 | |
| 7 | #include <QtCore/QDebug> |
| 8 | #include <QtCore/QCoreApplication> |
| 9 | #include <QtCore/QLoggingCategory> |
| 10 | |
| 11 | #include <libspeechd.h> |
| 12 | |
| 13 | #if LIBSPEECHD_MAJOR_VERSION > 0 || LIBSPEECHD_MINOR_VERSION >= 9 |
| 14 | #define HAVE_SPD_090 |
| 15 | #endif |
| 16 | |
| 17 | QT_BEGIN_NAMESPACE |
| 18 | |
| 19 | Q_LOGGING_CATEGORY(lcSpeechTtsSpeechd, "qt.speech.tts.speechd" ) |
| 20 | |
| 21 | typedef QList<QTextToSpeechEngineSpeechd*> QTextToSpeechSpeechDispatcherBackendList; |
| 22 | Q_GLOBAL_STATIC(QTextToSpeechSpeechDispatcherBackendList, backends) |
| 23 | |
| 24 | void speech_finished_callback(size_t msg_id, size_t client_id, SPDNotificationType state); |
| 25 | |
| 26 | QLocale QTextToSpeechEngineSpeechd::localeForVoice(SPDVoice *voice) const |
| 27 | { |
| 28 | QString lang_var = QString::fromLatin1(ba: voice->language); |
| 29 | if (qstrcmp(str1: voice->variant, str2: "none" ) != 0) { |
| 30 | QString var = QString::fromLatin1(ba: voice->variant); |
| 31 | lang_var += QLatin1Char('_') + var; |
| 32 | } |
| 33 | return QLocale(lang_var); |
| 34 | } |
| 35 | |
| 36 | QTextToSpeechEngineSpeechd::QTextToSpeechEngineSpeechd(const QVariantMap &, QObject *) |
| 37 | : speechDispatcher(nullptr) |
| 38 | { |
| 39 | backends->append(t: this); |
| 40 | connectToSpeechDispatcher(); |
| 41 | } |
| 42 | |
| 43 | QTextToSpeechEngineSpeechd::~QTextToSpeechEngineSpeechd() |
| 44 | { |
| 45 | if (speechDispatcher) { |
| 46 | if ((m_state != QTextToSpeech::Error) && (m_state != QTextToSpeech::Ready)) |
| 47 | spd_cancel_all(connection: speechDispatcher); |
| 48 | spd_close(connection: speechDispatcher); |
| 49 | } |
| 50 | backends->removeAll(t: this); |
| 51 | } |
| 52 | |
| 53 | bool QTextToSpeechEngineSpeechd::connectToSpeechDispatcher() |
| 54 | { |
| 55 | if (speechDispatcher) |
| 56 | return true; |
| 57 | |
| 58 | speechDispatcher = spd_open(client_name: "QTextToSpeech" , connection_name: "main" , user_name: nullptr, mode: SPD_MODE_THREADED); |
| 59 | if (!speechDispatcher) { |
| 60 | setError(reason: QTextToSpeech::ErrorReason::Initialization, |
| 61 | errorString: QCoreApplication::translate(context: "QTextToSpeech" , key: "Connection to speech-dispatcher failed" )); |
| 62 | return false; |
| 63 | } |
| 64 | |
| 65 | speechDispatcher->callback_begin = speech_finished_callback; |
| 66 | spd_set_notification_on(connection: speechDispatcher, notification: SPD_BEGIN); |
| 67 | speechDispatcher->callback_end = speech_finished_callback; |
| 68 | spd_set_notification_on(connection: speechDispatcher, notification: SPD_END); |
| 69 | speechDispatcher->callback_cancel = speech_finished_callback; |
| 70 | spd_set_notification_on(connection: speechDispatcher, notification: SPD_CANCEL); |
| 71 | speechDispatcher->callback_resume = speech_finished_callback; |
| 72 | spd_set_notification_on(connection: speechDispatcher, notification: SPD_RESUME); |
| 73 | speechDispatcher->callback_pause = speech_finished_callback; |
| 74 | spd_set_notification_on(connection: speechDispatcher, notification: SPD_PAUSE); |
| 75 | |
| 76 | QStringList availableModules; |
| 77 | char **modules = spd_list_modules(connection: speechDispatcher); |
| 78 | int i = 0; |
| 79 | while (modules && modules[i]) { |
| 80 | availableModules.append(t: QString::fromUtf8(utf8: modules[i])); |
| 81 | ++i; |
| 82 | } |
| 83 | |
| 84 | if (availableModules.length() == 0) { |
| 85 | setError(reason: QTextToSpeech::ErrorReason::Configuration, |
| 86 | errorString: QCoreApplication::translate(context: "QTextToSpeech" , |
| 87 | key: "Found no modules in speech-dispatcher." )); |
| 88 | return false; |
| 89 | } |
| 90 | |
| 91 | updateVoices(); |
| 92 | if (m_currentVoice == QVoice()) { |
| 93 | // Set the default locale (which is usually the system locale), and fall back |
| 94 | // to a locale that has the same language if that fails. That might then still fail, |
| 95 | // in which case there won't be a valid voice. |
| 96 | if (!setLocale(QLocale()) && !setLocale(QLocale().language())) { |
| 97 | setError(reason: QTextToSpeech::ErrorReason::Configuration, |
| 98 | errorString: QCoreApplication::translate(context: "QTextToSpeech" , |
| 99 | key: "Failed to initialize default locale and voice." )); |
| 100 | return false; |
| 101 | } |
| 102 | } |
| 103 | |
| 104 | m_state = QTextToSpeech::Ready; |
| 105 | m_errorReason = QTextToSpeech::ErrorReason::NoError; |
| 106 | m_errorString.clear(); |
| 107 | |
| 108 | return true; |
| 109 | } |
| 110 | |
| 111 | // hack to get state notifications |
| 112 | void QTextToSpeechEngineSpeechd::spdStateChanged(SPDNotificationType state) |
| 113 | { |
| 114 | QTextToSpeech::State s = QTextToSpeech::Error; |
| 115 | if (state == SPD_EVENT_PAUSE) |
| 116 | s = QTextToSpeech::Paused; |
| 117 | else if ((state == SPD_EVENT_BEGIN) || (state == SPD_EVENT_RESUME)) |
| 118 | s = QTextToSpeech::Speaking; |
| 119 | else if ((state == SPD_EVENT_CANCEL) || (state == SPD_EVENT_END)) |
| 120 | s = QTextToSpeech::Ready; |
| 121 | |
| 122 | if (m_state != s) { |
| 123 | m_state = s; |
| 124 | emit stateChanged(state: m_state); |
| 125 | } |
| 126 | } |
| 127 | |
| 128 | void QTextToSpeechEngineSpeechd::setError(QTextToSpeech::ErrorReason reason, const QString &errorString) |
| 129 | { |
| 130 | m_errorReason = reason; |
| 131 | m_errorString = errorString; |
| 132 | if (reason == QTextToSpeech::ErrorReason::NoError) { |
| 133 | m_errorString.clear(); |
| 134 | return; |
| 135 | } |
| 136 | |
| 137 | if (m_state != QTextToSpeech::Error) { |
| 138 | m_state = QTextToSpeech::Error; |
| 139 | emit stateChanged(state: m_state); |
| 140 | } |
| 141 | emit errorOccurred(error: m_errorReason, errorString: m_errorString); |
| 142 | } |
| 143 | |
| 144 | void QTextToSpeechEngineSpeechd::say(const QString &text) |
| 145 | { |
| 146 | if (text.isEmpty() || !connectToSpeechDispatcher()) |
| 147 | return; |
| 148 | |
| 149 | if (m_state != QTextToSpeech::Ready) |
| 150 | stop(boundaryHint: QTextToSpeech::BoundaryHint::Default); |
| 151 | |
| 152 | if (spd_say(connection: speechDispatcher, priority: SPD_MESSAGE, text: text.toUtf8().constData()) < 0) |
| 153 | setError(reason: QTextToSpeech::ErrorReason::Input, |
| 154 | errorString: QCoreApplication::translate(context: "QTextToSpeech" , key: "Text synthesizing failure." )); |
| 155 | } |
| 156 | |
| 157 | void QTextToSpeechEngineSpeechd::synthesize(const QString &) |
| 158 | { |
| 159 | setError(reason: QTextToSpeech::ErrorReason::Configuration, errorString: tr(s: "Synthesize not supported" )); |
| 160 | } |
| 161 | |
| 162 | void QTextToSpeechEngineSpeechd::stop(QTextToSpeech::BoundaryHint boundaryHint) |
| 163 | { |
| 164 | Q_UNUSED(boundaryHint); |
| 165 | if (!connectToSpeechDispatcher()) |
| 166 | return; |
| 167 | |
| 168 | if (m_state == QTextToSpeech::Paused) |
| 169 | spd_resume_all(connection: speechDispatcher); |
| 170 | spd_cancel_all(connection: speechDispatcher); |
| 171 | } |
| 172 | |
| 173 | void QTextToSpeechEngineSpeechd::pause(QTextToSpeech::BoundaryHint boundaryHint) |
| 174 | { |
| 175 | Q_UNUSED(boundaryHint); |
| 176 | if (!connectToSpeechDispatcher()) |
| 177 | return; |
| 178 | |
| 179 | if (m_state == QTextToSpeech::Speaking) { |
| 180 | spd_pause_all(connection: speechDispatcher); |
| 181 | } |
| 182 | } |
| 183 | |
| 184 | void QTextToSpeechEngineSpeechd::resume() |
| 185 | { |
| 186 | if (!connectToSpeechDispatcher()) |
| 187 | return; |
| 188 | |
| 189 | if (m_state == QTextToSpeech::Paused) { |
| 190 | spd_resume_all(connection: speechDispatcher); |
| 191 | } |
| 192 | } |
| 193 | |
| 194 | bool QTextToSpeechEngineSpeechd::setPitch(double pitch) |
| 195 | { |
| 196 | if (!connectToSpeechDispatcher()) |
| 197 | return false; |
| 198 | |
| 199 | int result = spd_set_voice_pitch(connection: speechDispatcher, pitch: static_cast<int>(pitch * 100)); |
| 200 | if (result == 0) |
| 201 | return true; |
| 202 | return false; |
| 203 | } |
| 204 | |
| 205 | double QTextToSpeechEngineSpeechd::pitch() const |
| 206 | { |
| 207 | double pitch = 0.0; |
| 208 | #ifdef HAVE_SPD_090 |
| 209 | if (speechDispatcher != 0) { |
| 210 | int result = spd_get_voice_pitch(connection: speechDispatcher); |
| 211 | pitch = result / 100.0; |
| 212 | } |
| 213 | #endif |
| 214 | return pitch; |
| 215 | } |
| 216 | |
| 217 | bool QTextToSpeechEngineSpeechd::setRate(double rate) |
| 218 | { |
| 219 | if (!connectToSpeechDispatcher()) |
| 220 | return false; |
| 221 | |
| 222 | int result = spd_set_voice_rate(connection: speechDispatcher, rate: static_cast<int>(rate * 100)); |
| 223 | return result == 0; |
| 224 | } |
| 225 | |
| 226 | double QTextToSpeechEngineSpeechd::rate() const |
| 227 | { |
| 228 | double rate = 0.0; |
| 229 | #ifdef HAVE_SPD_090 |
| 230 | if (speechDispatcher != 0) { |
| 231 | int result = spd_get_voice_rate(connection: speechDispatcher); |
| 232 | rate = result / 100.0; |
| 233 | } |
| 234 | #endif |
| 235 | return rate; |
| 236 | } |
| 237 | |
| 238 | bool QTextToSpeechEngineSpeechd::setVolume(double volume) |
| 239 | { |
| 240 | if (!connectToSpeechDispatcher()) |
| 241 | return false; |
| 242 | |
| 243 | // convert from 0.0..1.0 to -100..100 |
| 244 | int result = spd_set_volume(connection: speechDispatcher, volume: (volume - 0.5) * 200); |
| 245 | return result == 0; |
| 246 | } |
| 247 | |
| 248 | double QTextToSpeechEngineSpeechd::volume() const |
| 249 | { |
| 250 | double volume = 0.0; |
| 251 | #ifdef HAVE_SPD_090 |
| 252 | if (speechDispatcher != 0) { |
| 253 | int result = spd_get_volume(connection: speechDispatcher); |
| 254 | // -100..100 to 0.0..1.0 |
| 255 | volume = (result + 100) / 200.0; |
| 256 | } |
| 257 | #endif |
| 258 | return volume; |
| 259 | } |
| 260 | |
| 261 | bool QTextToSpeechEngineSpeechd::setLocale(const QLocale &locale) |
| 262 | { |
| 263 | if (!connectToSpeechDispatcher()) |
| 264 | return false; |
| 265 | |
| 266 | const int result = spd_set_language(connection: speechDispatcher, language: locale.uiLanguages().at(i: 0).toUtf8().data()); |
| 267 | if (result == 0) { |
| 268 | const QVoice previousVoice = m_currentVoice; |
| 269 | |
| 270 | const QList<QVoice> voices = m_voices.values(key: locale); |
| 271 | // QMultiHash returns the values in the reverse order |
| 272 | if (voices.size() > 0 && setVoice(voices.last())) |
| 273 | return true; |
| 274 | |
| 275 | // try to go back to the previous locale/voice |
| 276 | setVoice(previousVoice); |
| 277 | } |
| 278 | setError(reason: QTextToSpeech::ErrorReason::Configuration, |
| 279 | errorString: QCoreApplication::translate(context: "QTextToSpeech" , key: "Locale not available: %1" ) |
| 280 | .arg(a: locale.name())); |
| 281 | return false; |
| 282 | } |
| 283 | |
| 284 | QLocale QTextToSpeechEngineSpeechd::locale() const |
| 285 | { |
| 286 | return m_currentVoice.locale(); |
| 287 | } |
| 288 | |
| 289 | bool QTextToSpeechEngineSpeechd::setVoice(const QVoice &voice) |
| 290 | { |
| 291 | if (!connectToSpeechDispatcher()) |
| 292 | return false; |
| 293 | |
| 294 | const QByteArray moduleName = voiceData(voice).value<QByteArray>(); |
| 295 | const int result = spd_set_output_module(connection: speechDispatcher, output_module: moduleName); |
| 296 | if (result != 0) { |
| 297 | setError(reason: QTextToSpeech::ErrorReason::Configuration, |
| 298 | errorString: QCoreApplication::translate(context: "QTextToSpeech" , |
| 299 | key: "Output module %1, associated with voice %2 not available" ) |
| 300 | .arg(a: moduleName).arg(a: voice.name())); |
| 301 | return false; |
| 302 | } |
| 303 | const int result2 = spd_set_synthesis_voice(speechDispatcher, voice_name: voice.name().toUtf8().data()); |
| 304 | if (result2 == 0) { |
| 305 | m_currentVoice = voice; |
| 306 | return true; |
| 307 | } |
| 308 | setError(reason: QTextToSpeech::ErrorReason::Configuration, |
| 309 | errorString: QCoreApplication::translate(context: "QTextToSpeech" , key: "Invalid voice: %1" ) |
| 310 | .arg(a: voice.name())); |
| 311 | return false; |
| 312 | } |
| 313 | |
| 314 | QVoice QTextToSpeechEngineSpeechd::voice() const |
| 315 | { |
| 316 | return m_currentVoice; |
| 317 | } |
| 318 | |
| 319 | QTextToSpeech::State QTextToSpeechEngineSpeechd::state() const |
| 320 | { |
| 321 | return m_state; |
| 322 | } |
| 323 | |
| 324 | QTextToSpeech::ErrorReason QTextToSpeechEngineSpeechd::errorReason() const |
| 325 | { |
| 326 | return m_errorReason; |
| 327 | } |
| 328 | QString QTextToSpeechEngineSpeechd::errorString() const |
| 329 | { |
| 330 | return m_errorString; |
| 331 | } |
| 332 | |
| 333 | void QTextToSpeechEngineSpeechd::updateVoices() |
| 334 | { |
| 335 | char **modules = spd_list_modules(connection: speechDispatcher); |
| 336 | #ifdef HAVE_SPD_090 |
| 337 | char *original_module = spd_get_output_module(connection: speechDispatcher); |
| 338 | #else |
| 339 | char *original_module = modules[0]; |
| 340 | #endif |
| 341 | char **module = modules; |
| 342 | while (module != nullptr && module[0] != nullptr) { |
| 343 | spd_set_output_module(connection: speechDispatcher, output_module: module[0]); |
| 344 | |
| 345 | SPDVoice **voices = spd_list_synthesis_voices(connection: speechDispatcher); |
| 346 | int i = 0; |
| 347 | while (voices != nullptr && voices[i] != nullptr) { |
| 348 | const QLocale locale = localeForVoice(voice: voices[i]); |
| 349 | const QVariant data = QVariant::fromValue<QByteArray>(value: module[0]); |
| 350 | // speechd declares enums and APIs for gender and age, but the SPDVoice struct |
| 351 | // carries no relevant information. |
| 352 | const QVoice voice = createVoice(name: QString::fromUtf8(utf8: voices[i]->name), locale, |
| 353 | gender: QVoice::Unknown, age: QVoice::Other, data); |
| 354 | m_voices.insert(key: locale, value: voice); |
| 355 | ++i; |
| 356 | } |
| 357 | // free voices. |
| 358 | #ifdef HAVE_SPD_090 |
| 359 | free_spd_voices(voices); |
| 360 | #endif |
| 361 | ++module; |
| 362 | } |
| 363 | |
| 364 | #ifdef HAVE_SPD_090 |
| 365 | // Also free modules. |
| 366 | free_spd_modules(modules); |
| 367 | #endif |
| 368 | // Set the output module back to what it was. |
| 369 | spd_set_output_module(connection: speechDispatcher, output_module: original_module); |
| 370 | #ifdef HAVE_SPD_090 |
| 371 | free(ptr: original_module); |
| 372 | #endif |
| 373 | } |
| 374 | |
| 375 | QList<QLocale> QTextToSpeechEngineSpeechd::availableLocales() const |
| 376 | { |
| 377 | return m_voices.uniqueKeys(); |
| 378 | } |
| 379 | |
| 380 | QList<QVoice> QTextToSpeechEngineSpeechd::availableVoices() const |
| 381 | { |
| 382 | QList<QVoice> resultList = m_voices.values(key: m_currentVoice.locale()); |
| 383 | std::reverse(first: resultList.begin(), last: resultList.end()); |
| 384 | return resultList; |
| 385 | } |
| 386 | |
| 387 | // We have no way of knowing our own client_id since speech-dispatcher seems to be incomplete |
| 388 | // (history functions are just stubs) |
| 389 | void speech_finished_callback(size_t msg_id, size_t client_id, SPDNotificationType state) |
| 390 | { |
| 391 | qCDebug(lcSpeechTtsSpeechd) << "Message from speech dispatcher" << msg_id << client_id; |
| 392 | for (QTextToSpeechEngineSpeechd *backend : std::as_const(t&: *backends)) |
| 393 | backend->spdStateChanged(state); |
| 394 | } |
| 395 | |
| 396 | QT_END_NAMESPACE |
| 397 | |