// Initializes the TTS engine and returns whether initialization succeeded
tts_result TtsEngine::init(synthDoneCB_t synthDoneCBPtr)
{
// TODO Make sure that the speech data is loaded in
// the directory /sdcard/espeak-data before calling this.
int sampleRate = espeak_Initialize(AUDIO_OUTPUT_SYNCHRONOUS,
4096, "/sdcard", 0);
if (sampleRate <= 0) {
LOGI("eSpeak initialization failed!");
return TTS_FAILURE;
}
espeak_SetSynthCallback(eSpeakCallback);
int speechRate = 140;
espeak_ERROR err = espeak_SetParameter(espeakRATE, speechRate, 0);
espeak_VOICE voice;
memset( &voice, 0, sizeof(espeak_VOICE)); // Zero out the voice first
const char *langNativeString = "en-us"; //Default to US English
voice.languages = langNativeString;
voice.variant = 0;
err = espeak_SetVoiceByProperties(&voice);
ttsSynthDoneCBPointer = synthDoneCBPtr;
return TTS_SUCCESS;
}
JNIEXPORT jboolean
JNICALL Java_com_googlecode_eyesfree_espeak_SpeechSynthesis_nativeSetVoiceByProperties(
JNIEnv *env, jobject object, jstring name, jstring languages, jint gender, jint age,
jint variant) {
if (DEBUG) LOGV("%s", __FUNCTION__);
const char *c_name = env->GetStringUTFChars(name, NULL);
const char *c_languages = env->GetStringUTFChars(languages, NULL);
espeak_VOICE voice_select;
memset(&voice_select, 0, sizeof(espeak_VOICE));
voice_select.name = c_name;
voice_select.languages = c_languages;
voice_select.age = (int) age;
voice_select.gender = (int) gender;
voice_select.variant = (int) variant;
const espeak_ERROR result = espeak_SetVoiceByProperties(&voice_select);
env->ReleaseStringUTFChars(name, c_name);
env->ReleaseStringUTFChars(languages, c_languages);
if (result == EE_OK)
return JNI_TRUE;
else
return JNI_FALSE;
}
void QGCAudioWorker::init()
{
#if defined Q_OS_LINUX && !defined __android__ && defined QGC_SPEECH_ENABLED
espeak_Initialize(AUDIO_OUTPUT_SYNCH_PLAYBACK, 500, NULL, 0); // initialize for playback with 500ms buffer and no options (see speak_lib.h)
espeak_VOICE *espeak_voice = espeak_GetCurrentVoice();
espeak_voice->languages = "en-uk"; // Default to British English
espeak_voice->identifier = NULL; // no specific voice file specified
espeak_voice->name = "klatt"; // espeak voice name
espeak_voice->gender = 2; // Female
espeak_voice->age = 0; // age not specified
espeak_SetVoiceByProperties(espeak_voice);
#endif
#if defined _MSC_VER && defined QGC_SPEECH_ENABLED
if (FAILED(::CoInitialize(NULL)))
{
qDebug() << "ERROR: Creating COM object for audio output failed!";
}
else
{
HRESULT hr = CoCreateInstance(CLSID_SpVoice, NULL, CLSCTX_ALL, IID_ISpVoice, (void **)&pVoice);
if (FAILED(hr))
{
qDebug() << "ERROR: Initializing voice for audio output failed!";
}
}
#endif
}
long set_voice(
const char* aName,
const char* aLang=NULL,
unsigned char aGender=0,
unsigned char aAge=0,
unsigned char aVariant = 0
) {
long result = 0;
if (aLang || aGender || aAge || aVariant) {
espeak_VOICE props = { 0 };
props.name = aName;
props.languages = aLang;
props.gender = aGender;
props.age = aAge;
props.variant = aVariant;
result = espeak_SetVoiceByProperties(&props);
} else {
result = espeak_SetVoiceByName(aName);
}
// This way we don't need to allocate the name/lang strings to the heap.
// Instead, we store the actual global voice.
current_voice = espeak_GetCurrentVoice();
return result;
}
static int spk_construct(volatile SpeechSynthesizer *spk, char **parameters)
{
char *data_path, *voicename, *punctlist;
int result;
spk->setVolume = spk_setVolume;
spk->setRate = spk_setRate;
spk->setPitch = spk_setPitch;
spk->setPunctuation = spk_setPunctuation;
spk->drain = spk_drain;
logMessage(LOG_INFO, "eSpeak-NG version %s", espeak_Info(NULL));
data_path = parameters[PARM_PATH];
if (data_path && !*data_path)
data_path = NULL;
result = espeak_Initialize(AUDIO_OUTPUT_PLAYBACK, 0, data_path, 0);
if (result < 0) {
logMessage(LOG_ERR, "eSpeak-NG: initialization failed");
return 0;
}
voicename = parameters[PARM_VOICE];
if(!voicename || !*voicename)
voicename = "en";
result = espeak_SetVoiceByName(voicename);
if (result != EE_OK) {
espeak_VOICE voice_select;
memset(&voice_select, 0, sizeof(voice_select));
voice_select.languages = voicename;
result = espeak_SetVoiceByProperties(&voice_select);
}
if (result != EE_OK) {
logMessage(LOG_ERR, "eSpeak-NG: unable to load voice '%s'", voicename);
return 0;
}
punctlist = parameters[PARM_PUNCTLIST];
if (punctlist && *punctlist) {
wchar_t w_punctlist[strlen(punctlist) + 1];
int i = 0;
while ((w_punctlist[i] = punctlist[i]) != 0) i++;
espeak_SetPunctuationList(w_punctlist);
}
if (parameters[PARM_MAXRATE]) {
int val = atoi(parameters[PARM_MAXRATE]);
if (val > espeakRATE_MINIMUM) maxrate = val;
}
espeak_SetSynthCallback(SynthCallback);
return 1;
}
void dyn_cfg_callback(espeak_ros::EspeakConfig &cfg, uint32_t level) {
// lock mutex before calling espeak functions
boost::mutex::scoped_lock u_lock(mtx);
std::string voice, dialect;
voice = getVoiceName(cfg.voice);
dialect = getDialectName(cfg.dialect);
espeak_VOICE voice_select;
std::memset(&voice_select,0,sizeof(voice_select));
voice_select.name = voice.c_str();
voice_select.languages = dialect.c_str();
voice_select.age = cfg.age; // 0=not specified, or age in years
voice_select.gender = cfg.gender; // 0=none, 1=male, 2=female
if (espeak_SetVoiceByProperties(&voice_select) != EE_OK) {
ROS_ERROR("Could not set espeak voice properties. Aborting.");
return;
}
ROS_INFO("Using voice %s", voice.c_str());
ROS_INFO("Using dialect %s", dialect.c_str());
if (espeak_SetParameter(espeakRATE, cfg.rate, 0) != EE_OK) {
ROS_ERROR("Could not set espeak rate. Aborting.");
return;
}
if (espeak_SetParameter(espeakVOLUME, cfg.volume, 0) != EE_OK) {
ROS_ERROR("Could not set espeak volume. Aborting.");
return;
}
if (espeak_SetParameter(espeakPITCH, cfg.pitch, 0) != EE_OK) {
ROS_ERROR("Could not set espeak pitch. Aborting.");
return;
}
if (espeak_SetParameter(espeakRANGE, cfg.range, 0) != EE_OK) {
ROS_ERROR("Could not set espeak range. Aborting.");
return;
}
if (espeak_SetParameter(espeakPUNCTUATION,
espeak_PUNCT_TYPE(cfg.punctuation), 0) != EE_OK) {
ROS_ERROR("Could not set espeak punctuation. Aborting.");
return;
}
if (espeak_SetParameter(espeakCAPITALS, cfg.capitals, 0) != EE_OK) {
ROS_ERROR("Could not set espeak capitals. Aborting.");
return;
}
int wordgap = cfg.wordgap % 10;
if (espeak_SetParameter(espeakWORDGAP, wordgap, 0) != EE_OK) {
ROS_ERROR("Could not set espeak wordgap. Aborting.");
return;
}
}
static espeak_ERROR set_voice(struct synth_t *s, char *voice)
{
espeak_ERROR rc;
espeak_VOICE voice_select;
rc = espeak_SetVoiceByName(voice);
if (rc != EE_OK)
{
memset(&voice_select, 0, sizeof(voice_select));
voice_select.languages = voice;
rc = espeak_SetVoiceByProperties(&voice_select);
}
if (rc == EE_OK)
strcpy(s->voice, voice);
return rc;
}
void synth_(const char* aText, void* aCallback) {
t_espeak_callback* cb = reinterpret_cast<t_espeak_callback*>(aCallback);
espeak_SetSynthCallback(cb);
espeak_SetParameter(espeakPITCH, pitch, 0);
espeak_SetParameter(espeakRATE, rate, 0);
if (current_voice)
espeak_SetVoiceByProperties(current_voice);
else
espeak_SetVoiceByName("default");
espeak_Synth(aText, 0, 0, POS_CHARACTER, 0, 0, NULL, NULL);
// Reset callback so other instances will work too.
espeak_SetSynthCallback(NULL);
}
//-----------------------------------------------------------------------------
// InitSpeak - This function initializes the speech using the current file name
// or parameters...
//-----------------------------------------------------------------------------
void RobotSpeak::InitSpeak()
{
espeak_ERROR err = EE_NOT_FOUND; // something other than OK...
// See if this is the first time through if so register our release function.
if (_bIDSDRSpeak == 0xff)
_bIDSDRSpeak = SDRRegisterReleaseFunction(&_EndSpeak);
// Tell our quick and dirty Sound registry stuff, that we want to be the current one...
SDRSetCurrent(_bIDSDRSpeak);
esoutput = AUDIO_OUTPUT_PLAYBACK;
espeak_Initialize(esoutput, Buflength, path, Options );
// First see if someone registered a voice file name
if (_pszVoiceFileName)
{
err = espeak_SetVoiceByName(_pszVoiceFileName);
if (err != EE_OK)
printf("Speak SetVoiceByProp failed %d\n", err);
}
else if (_pszVoiceLanguage)
{
espeak_VOICE voice;
voice.name = NULL;
voice.languages = _pszVoiceLanguage;
voice.identifier= NULL;
voice.age = 0;
voice.gender = _bVoiceGender;
voice.variant = 0;
err = espeak_SetVoiceByProperties(&voice);
if (err != EE_OK)
printf("Speak SetVoiceByProp failed %d\n", err);
}
if (err != EE_OK)
espeak_SetVoiceByName(g_szDefaultVoice);
_fSpeakInit = true;
}
JNIEXPORT jboolean
JNICALL Java_com_googlecode_eyesfree_espeak_SpeechSynthesis_nativeSetLanguage(
JNIEnv *env, jobject object, jstring language, jint variant) {
if (DEBUG) LOGV("%s", __FUNCTION__);
const char *c_language = env->GetStringUTFChars(language, NULL);
const int len = strlen(c_language);
char *lang_copy = (char *) calloc(len, sizeof(char));
strcpy(lang_copy, c_language);
env->ReleaseStringUTFChars(language, c_language);
espeak_VOICE voice;
memset(&voice, 0, sizeof(espeak_VOICE)); // Zero out the voice first
voice.languages = lang_copy;
voice.variant = (int) variant;
const espeak_ERROR result = espeak_SetVoiceByProperties(&voice);
if (result == EE_OK)
return JNI_TRUE;
else
return JNI_FALSE;
}
int main(int argc, char* argv[] )
{
output = AUDIO_OUTPUT_PLAYBACK;
int I, Run = 1, L;
espeak_Initialize(output, Buflength, path, Options );
//espeak_SetVoiceByName(Voice);
const char *langNativeString = "lt"; //Default to US English
espeak_VOICE voice;
memset(&voice, 0, sizeof(espeak_VOICE)); // Zero out the voice first
voice.languages = langNativeString;
voice.name = "klatt";
voice.variant = 2;
voice.gender = 1;
espeak_SetVoiceByProperties(&voice);
Size = strlen(text)+1;
printf("Saying '%s'",text);
espeak_Synth( text, Size, position, position_type, end_position, flags,
unique_identifier, user_data );
espeak_Synchronize( );
printf("\n:Done\n");
return 0;
}
ESpeak::ESpeak(QObject *parent) :
QObject(parent),
queueLength(0)
{
if (0 == refcount)
{
espeak_Initialize(AUDIO_OUTPUT_PLAYBACK, 100, 0, 0);
espeak_VOICE voice;
QString language = QLocale::system().languageToString(QLocale::system().language());
QByteArray languageData = language.toAscii();
languageData.prepend((char)0);
languageData.append((char)0);
QByteArray languages = QByteArray(1, (char)0);
languageData += languages;
voice.languages = languageData.data();
voice.name = 0;
voice.gender = 2;
voice.age = 20;
espeak_SetVoiceByProperties(&voice);
espeak_SetParameter(espeakRATE, 175, false);
}
++refcount;
}
// Language will be specified according to the Android conventions for
// localization as documented here:
// http://developer.android.com/guide/topics/resources/resources-i18n.html
//
// language will be a string of the form "xx" or "xx-rYY", where xx is a
// two letter ISO 639-1 language code in lowercase and rYY is a two letter ISO
// 3166-1-alpha-2 language code in uppercase preceded by a lowercase "r".
// Note that the "-rYY" portion may be omitted if the region is unimportant.
//
tts_result TtsEngine::setLanguage( const char * lang, const char * country, const char * variant )
{
LOGE("lang input param: %s country input param: %s", lang, country);
char language[10];
int langIndex = -1;
for (int i = 0; i < languageCount; i ++)
{
if (strcmp(lang, supportedLangIso3[i]) == 0)
{
langIndex = i;
break;
}
}
if (langIndex < 0)
{
/* The language isn't supported. */
LOGE("TtsEngine::setLanguage called with unsupported language");
return TTS_FAILURE;
}
strcpy(currentLang, lang);
strcpy(currentCountry, country);
strcpy(language, supportedLang[langIndex]);
if (strcmp(language, "en") == 0){
if (strcmp(country, "USA") == 0){
strcpy(language, "en-rUS");
}
if (strcmp(country, "GBR") == 0){
strcpy(language, "en-rGB");
}
}
if (strcmp(language, "es") == 0){
if (strcmp(country, "MEX") == 0){
strcpy(language, "es-rMX");
}
}
LOGE("Language: %s", language);
espeak_VOICE voice;
memset(&voice, 0, sizeof(espeak_VOICE)); // Zero out the voice first
voice.variant = 0;
char espeakLangStr[6];
if ((strlen(language) != 2) && (strlen(language) != 6)){
LOGI("Error: Invalid language. Language must be in either xx or xx-rYY format.");
return TTS_VALUE_INVALID;
}
if (strcmp(language, "en-rUS") == 0){
strcpy(espeakLangStr, "en-us");
} else if (strcmp(language, "en-rGB") == 0){
strcpy(espeakLangStr, "en-uk");
} else if (strcmp(language, "es-rMX") == 0){
strcpy(espeakLangStr, "es-la");
} else if (strcmp(language, "zh-rHK") == 0){
strcpy(espeakLangStr, "zh");
voice.variant = 5;
} else {
espeakLangStr[0] = language[0];
espeakLangStr[1] = language[1];
espeakLangStr[2] = 0;
// Bail out and do nothing if the language is not supported by eSpeak
if ((strcmp(language, "af") != 0) &&
(strcmp(language, "bs") != 0) &&
(strcmp(language, "zh") != 0) &&
(strcmp(language, "hr") != 0) &&
(strcmp(language, "cz") != 0) &&
(strcmp(language, "nl") != 0) &&
(strcmp(language, "en") != 0) &&
(strcmp(language, "eo") != 0) &&
(strcmp(language, "fi") != 0) &&
(strcmp(language, "fr") != 0) &&
(strcmp(language, "de") != 0) &&
(strcmp(language, "el") != 0) &&
(strcmp(language, "hi") != 0) &&
(strcmp(language, "hu") != 0) &&
(strcmp(language, "is") != 0) &&
(strcmp(language, "id") != 0) &&
(strcmp(language, "it") != 0) &&
(strcmp(language, "ku") != 0) &&
(strcmp(language, "la") != 0) &&
(strcmp(language, "mk") != 0) &&
(strcmp(language, "no") != 0) &&
(strcmp(language, "pl") != 0) &&
(strcmp(language, "pt") != 0) &&
(strcmp(language, "ro") != 0) &&
(strcmp(language, "ru") != 0) &&
(strcmp(language, "sr") != 0) &&
(strcmp(language, "sk") != 0) &&
(strcmp(language, "es") != 0) &&
(strcmp(language, "sw") != 0) &&
(strcmp(language, "sv") != 0) &&
(strcmp(language, "ta") != 0) &&
(strcmp(language, "tr") != 0) &&
(strcmp(language, "vi") != 0) &&
(strcmp(language, "cy") != 0) ){
LOGI("Error: Unsupported language.");
return TTS_PROPERTY_UNSUPPORTED;
}
// Use American English as the default English
if (strcmp(language, "en") == 0) {
strcpy(espeakLangStr, "en-us");
}
}
voice.languages = espeakLangStr;
espeak_ERROR err = espeak_SetVoiceByProperties(&voice);
currentLanguage = new char [strlen(language)];
strcpy(currentLanguage, language);
return TTS_SUCCESS;
}
int main( int argc, char** argv ) {
espeak_Initialize(AUDIO_OUTPUT_PLAYBACK, 0, NULL, 0);
ros::init(argc, argv, "espeak_node");
ros::NodeHandle n;
ros::NodeHandle priv_n("~");
// for list of voices, run "espeak --voices"
// voices: "brazil", "default", "english", "lancashire",
// "english-rp", "english-wmids", "english-us", "en-scottish"
std::string voice_name;
// dialects: "pt-br", "en", "en-uk", "en-uk-north",
// "en-uk-rp", "en-uk-wmids", "en-us", "en-sc"
std::string dialect;
espeak_VOICE voice_select;
// See speak_lib.h for details and range of the parameters below
int rate, volume, pitch, range, punctuation;
int voice_num, dialect_num, capitals, wordgap, age, gender;
// Fetch default parameters
rate = espeak_GetParameter(espeakRATE, 0);
volume = espeak_GetParameter(espeakVOLUME, 0);
pitch = espeak_GetParameter(espeakPITCH, 0);
range = espeak_GetParameter(espeakRANGE, 0);
punctuation = espeak_GetParameter(espeakPUNCTUATION, 0);
capitals = espeak_GetParameter(espeakCAPITALS, 0);
wordgap = espeak_GetParameter(espeakWORDGAP, 0);
voice_num = 1;
dialect_num = 1;
// Fetch (and set if param could not be retrieved) ROS parameters
if (priv_n.getParam("voice", voice_num) == false) {
priv_n.setParam("voice", voice_num);
}
if (priv_n.getParam("dialect", dialect_num) == false) {
priv_n.setParam("dialect", dialect_num);
}
if (priv_n.getParam("rate", rate) == false) {
priv_n.setParam("rate", rate);
}
if (priv_n.getParam("volume", volume) == false) {
priv_n.setParam("volume", volume);
}
if (priv_n.getParam("pitch", pitch) == false) {
priv_n.setParam("pitch", pitch);
}
if (priv_n.getParam("punctuation", punctuation) == false) {
priv_n.setParam("punctuation", punctuation);
}
if (priv_n.getParam("capitals", capitals) == false) {
priv_n.setParam("capitals", capitals);
}
if (priv_n.getParam("wordgap", wordgap) == false) {
priv_n.setParam("wordgap", wordgap);
}
priv_n.param("age", age, int(0));
priv_n.param("gender", gender, int(2));
if (age < 0 || age > 100) {
age = 0;
}
if (gender < 0 || gender > 2) {
gender = 2;
}
voice_name = getVoiceName(voice_num);
dialect = getDialectName(dialect_num);
std::memset(&voice_select,0,sizeof(voice_select));
voice_select.name = voice_name.c_str();
voice_select.languages = dialect.c_str();
voice_select.age = age; // 0=not specified, or age in years
voice_select.gender = gender; // 0=none, 1=male, 2=female
// set espeak parameters
//if (espeak_SetVoiceByName(voice_name.c_str()) != EE_OK) {
// ROS_ERROR("Could not set espeak voice. Aborting.");
// return 1;
//}
if (espeak_SetVoiceByProperties(&voice_select) != EE_OK) {
ROS_ERROR("Could not set espeak voice properties. Aborting.");
return 1;
}
//ROS_INFO("Using voice %s", voice_name.c_str());
//ROS_INFO("Using dialect %s", dialect.c_str());
if (rate < 80 || rate > 450) {
ROS_INFO("Parameter rate = %d out of range, using default value", rate);
}
else {
if (espeak_SetParameter(espeakRATE, rate, 0) != EE_OK) {
ROS_ERROR("Could not set espeak rate. Aborting.");
return 1;
}
}
if (volume < 0 || volume > 200) {
ROS_INFO("Parameter volume = %d out of range, using default value", volume);
}
else {
if (espeak_SetParameter(espeakVOLUME, volume, 0) != EE_OK) {
ROS_ERROR("Could not set espeak volume. Aborting.");
return 1;
}
}
if (pitch < 0 || pitch > 100) {
ROS_INFO("Parameter pitch = %d out of range, using default value", pitch);
}
else {
if (espeak_SetParameter(espeakPITCH, pitch, 0) != EE_OK) {
ROS_ERROR("Could not set espeak pitch. Aborting.");
return 1;
}
}
if (range < 0 || range > 100) {
ROS_INFO("Parameter range = %d out of range, using default value", range);
}
else {
if (espeak_SetParameter(espeakRANGE, range, 0) != EE_OK) {
ROS_ERROR("Could not set espeak range. Aborting.");
return 1;
}
}
if (punctuation < 0 || punctuation > 2) {
ROS_INFO("Parameter punctuation out of range, using default value");
}
else {
if (espeak_SetParameter(espeakPUNCTUATION,
espeak_PUNCT_TYPE(punctuation), 0) != EE_OK) {
ROS_ERROR("Could not set espeak punctuation. Aborting.");
return 1;
}
}
if (capitals < 0 || capitals > 3) {
ROS_INFO("Parameter capitals out of range, using default value");
}
else {
if (espeak_SetParameter(espeakCAPITALS, capitals, 0) != EE_OK) {
ROS_ERROR("Could not set espeak capitals. Aborting.");
return 1;
}
}
if (wordgap < 0 || wordgap > 1000 || wordgap % 10 != 0) {
ROS_INFO("Parameter wordgap out of range, using default value");
}
else {
if (espeak_SetParameter(espeakWORDGAP, wordgap, 0) != EE_OK) {
ROS_ERROR("Could not set espeak wordgap. Aborting.");
return 1;
}
}
dynamic_reconfigure::Server<espeak_ros::EspeakConfig> server;
dynamic_reconfigure::Server<espeak_ros::EspeakConfig>::CallbackType f;
f = boost::bind(&dyn_cfg_callback, _1, _2);
server.setCallback(f);
ros::Subscriber sub = n.subscribe("/espeak_node/speak_line", 20, espeak_callback);
ros::spin();
while (n.ok());
return 0;
}
