//请求指定语言的语音
bool SpBox::RequireLanguage(LPCWSTR langid)
{
IEnumSpObjectTokens* pEnum; //语音总选择
ISpObjectToken* pVoiceToken; //当前语音选择
ULONG ulCount = 0; //选择个数
WCHAR* pVoiceLang; //语音LangId
WCHAR* pVoiceText; //语音说明文字
HRESULT hr = SpEnumTokens(SPCAT_VOICES, NULL, NULL, &pEnum);
if (SUCCEEDED(hr))
{
hr = pEnum->GetCount(&ulCount);
}
while (SUCCEEDED(hr) && ulCount--){
if (SUCCEEDED(hr))
pEnum->Next(1, &pVoiceToken, 0);
if (SUCCEEDED(hr))
hr = SpGetLanguage(pVoiceToken, &pVoiceLang);
if (SUCCEEDED(hr) && !_wcsnicmp(pVoiceLang, langid, 3))
{
Token = pVoiceToken;
VoiceObj->SetVoice(Token);
SpGetDescription(Token, &pVoiceText, 0);
Description = pVoiceText;
return true;
}
}
return false;
}
bool EnumerateVoices() {
HRESULT comResult = S_OK;
ISpObjectTokenCategory * comTokenCategory = NULL;
IEnumSpObjectTokens * comVoices = NULL;
ULONG comVoicesCount = 0;
// Init speech api
comResult = ::CoCreateInstance(
CLSID_SpVoice, NULL, CLSCTX_INPROC_SERVER, IID_ISpVoice, (LPVOID*)&comVoice);
wxCHECK_MSG( SUCCEEDED(comResult), false, _T("Unable to instantiate speech API"));
// Generate enumeration of voices
comResult = ::CoCreateInstance(CLSID_SpObjectTokenCategory, NULL,
CLSCTX_INPROC_SERVER, IID_ISpObjectTokenCategory, (LPVOID*)&comTokenCategory);
wxCHECK_MSG( SUCCEEDED(comResult), false, _T("Unable to instantiate a TokenCategory"));
comResult = comTokenCategory->SetId(SPCAT_VOICES, false);
wxCHECK_MSG( SUCCEEDED(comResult), false,
_T("Unable to to set location to find the installed voices.")
_T("Likely the key that I am looking for does not exist and thus ")
_T("there are no installed voices on this system."));
comResult = comTokenCategory->EnumTokens(NULL, NULL, &comVoices);
wxCHECK_MSG( SUCCEEDED(comResult), false,
_T("Unable to enumerate the installed voices. Check that this system")
_T(" has voices installed."));
comResult = comVoices->GetCount(&comVoicesCount);
wxCHECK_MSG( SUCCEEDED(comResult), false,
_T("Unable get a count of the installed voices."));
while( comVoicesCount > 0 ) {
ISpObjectToken * comAVoice = NULL;
comVoices->Next(1, &comAVoice, NULL); // retrieve just one
LPWSTR id = NULL;
comAVoice->GetStringValue(NULL, &id);
size_t idlength = wcslen(id);
wxLogDebug(_T(" Got string of length %ld:"), idlength);
for(size_t i = 0; i < idlength; i++) {
wxLogDebug(_T(" %04X"), id[i]);
}
voices.push_back(VoiceData(wxString(id, wxMBConvUTF16(), wcslen(id)), comAVoice));
#ifdef __WXDEBUG__
enumerateObjectToken(comAVoice);
#endif
comAVoice->Release();
comVoicesCount--;
}
comTokenCategory->Release();
return true;
}
void Speech::selectOutputDevice() {
HRESULT hr = S_OK;
ISpObjectToken* cpAudioOutToken;
IEnumSpObjectTokens* cpEnum;
ULONG ulCount = 0;
if (FAILED(CoInitialize(NULL)))
return;
if (SUCCEEDED (hr)) {
hr = SpEnumTokens( SPCAT_AUDIOOUT, NULL, NULL, &cpEnum);
}
if (SUCCEEDED (hr)) {
hr = cpEnum->GetCount( &ulCount);
}
std::cout << "Select Audio Device:" << std::endl;
unsigned int i=0;
vector<wchar_t*> deviceIds;
while (SUCCEEDED(hr) && i < ulCount) {
i++;
if (SUCCEEDED (hr)) {
hr = cpEnum->Next( 1, &cpAudioOutToken, NULL );
}
wchar_t* deviceName = nullptr;
SpGetDescription(cpAudioOutToken, &deviceName);
wchar_t* deviceId = nullptr;
hr = cpAudioOutToken->GetStringValue(L"DeviceId", &deviceId);
deviceIds.push_back(deviceId);
std::wcout << "(" << i << ") " << deviceName << std::endl;
}
cpAudioOutToken->Release();
cpEnum->Release();
CoUninitialize();
i=5;
while(i < 1 || i > ulCount) {
cin >> i;
if(i < 1 || i > ulCount) {
cout << "Invalid Input." << endl;
cin.clear();
cin.ignore(10000, '\n');
}
}
outputDevice = deviceIds.at(i-1);
return;
}
void Speech::speechOutput(const char* str) {
ISpVoice* pVoice = NULL;
ISpObjectToken* cpAudioOutToken;
IEnumSpObjectTokens* cpEnum;
ULONG ulCount = 0;
if (FAILED(::CoInitialize(NULL)))
return;
HRESULT hr = CoCreateInstance(CLSID_SpVoice, NULL, CLSCTX_ALL, IID_ISpVoice, (void **)&pVoice);
if (SUCCEEDED (hr)) {
hr = SpEnumTokens( SPCAT_AUDIOOUT, NULL, NULL, &cpEnum);
}
if (SUCCEEDED (hr)) {
hr = cpEnum->GetCount( &ulCount);
}
while (SUCCEEDED(hr) && ulCount--) {
hr = cpEnum->Next( 1, &cpAudioOutToken, NULL );
wchar_t* deviceId = nullptr;
if(SUCCEEDED(hr)) {
hr = cpAudioOutToken->GetStringValue(L"DeviceId", &deviceId);
if(wcscmp(outputDevice, deviceId) == 0) break;
}
}
if (SUCCEEDED (hr)) {
hr = pVoice->SetOutput( cpAudioOutToken, TRUE );
}
if(SUCCEEDED(hr)) {
const size_t cSize = strlen(str)+1;
wchar_t* wc = new wchar_t[cSize];
size_t ret;
mbstowcs_s (&ret, wc, cSize, str, cSize);
hr = pVoice->Speak(wc, NULL, NULL);
pVoice->Release();
pVoice = NULL;
}
cpEnum->Release();
cpAudioOutToken->Release();
::CoUninitialize();
}
void DumpCategory(LPCWSTR category) {
// enumerate tokens in each category
IEnumSpObjectTokens *pEnumSpObjectTokens = nullptr;
HRESULT hr = SpEnumTokens(category, nullptr, nullptr, &pEnumSpObjectTokens);
if (SPERR_NOT_FOUND == hr) {
LOG(L" None found.");
return;
} else if (FAILED(hr)) {
ERR(L"SpEnumTokens failed: hr = 0x%08x", hr);
return;
}
ReleaseOnExit rEnumSpObjectTokens(pEnumSpObjectTokens);
ULONG nTokens = 0;
hr = pEnumSpObjectTokens->GetCount(&nTokens);
if (FAILED(hr)) {
ERR(L"IEnumSpObjectTokens::GetCount failed: hr = 0x%08x", hr);
return;
}
for (ULONG token = 0; token < nTokens; token++) {
ISpObjectToken *pSpObjectToken = nullptr;
hr = pEnumSpObjectTokens->Next(1, &pSpObjectToken, nullptr);
if (FAILED(hr)) {
ERR(L"IEnumSpObjectTokens::Next failed: hr = 0x%08x", hr);
return;
}
ReleaseOnExit rSpObjectToken(pSpObjectToken);
LPWSTR description = nullptr;
hr = SpGetDescription(pSpObjectToken, &description);
if (FAILED(hr)) {
ERR(L"SpGetDescription failed: hr = 0x%08x", hr);
continue;
}
CoTaskMemFreeOnExit fDescription(description);
LOG(L" #%u: %s", token + 1, description);
EnumDataKey(2, pSpObjectToken);
}
}
SCP_vector<SCP_string> speech_enumerate_voices()
{
#ifdef _WIN32
HRESULT hr = CoCreateInstance(
CLSID_SpVoice,
NULL,
CLSCTX_ALL,
IID_ISpVoice,
(void **)&Voice_device);
if (FAILED(hr)) {
return SCP_vector<SCP_string>();
}
// This code is mostly copied from wxLauncher
ISpObjectTokenCategory * comTokenCategory = NULL;
IEnumSpObjectTokens * comVoices = NULL;
ULONG comVoicesCount = 0;
// Generate enumeration of voices
hr = ::CoCreateInstance(CLSID_SpObjectTokenCategory, NULL,
CLSCTX_INPROC_SERVER, IID_ISpObjectTokenCategory, (LPVOID*)&comTokenCategory);
if (FAILED(hr)) {
return SCP_vector<SCP_string>();
}
hr = comTokenCategory->SetId(SPCAT_VOICES, false);
if (FAILED(hr)) {
return SCP_vector<SCP_string>();
}
hr = comTokenCategory->EnumTokens(NULL, NULL, &comVoices);
if (FAILED(hr)) {
return SCP_vector<SCP_string>();
}
hr = comVoices->GetCount(&comVoicesCount);
if (FAILED(hr)) {
return SCP_vector<SCP_string>();
}
SCP_vector<SCP_string> voices;
while (comVoicesCount > 0) {
ISpObjectToken * comAVoice = NULL;
comVoices->Next(1, &comAVoice, NULL); // retrieve just one
LPWSTR id = NULL;
comAVoice->GetStringValue(NULL, &id);
auto idlength = wcslen(id);
auto buffer_size = WideCharToMultiByte(CP_UTF8, 0, id, (int)idlength, nullptr, 0, nullptr, nullptr);
if (buffer_size > 0) {
SCP_string voiceName;
voiceName.resize(buffer_size);
buffer_size = WideCharToMultiByte(CP_UTF8, 0, id, (int)idlength, &voiceName[0], buffer_size, nullptr, nullptr);
voices.push_back(voiceName);
}
CoTaskMemFree(id);
comAVoice->Release();
comVoicesCount--;
}
comTokenCategory->Release();
Voice_device->Release();
return voices;
#else
STUB_FUNCTION;
return SCP_vector<SCP_string>();
#endif
}
static int Create (vlc_object_t *p_this)
{
filter_t *p_filter = (filter_t *)p_this;
filter_sys_t *p_sys;
HRESULT hr;
p_filter->p_sys = p_sys = (filter_sys_t*) malloc(sizeof(filter_sys_t));
if (!p_sys)
return VLC_ENOMEM;
if (TryEnterMTA(p_this))
goto error;
p_sys->cpVoice = nullptr;
hr = CoCreateInstance(CLSID_SpVoice, nullptr, CLSCTX_INPROC_SERVER, IID_ISpVoice, (void**) &p_sys->cpVoice);
if (SUCCEEDED(hr)) {
ISpObjectToken* cpVoiceToken = nullptr;
IEnumSpObjectTokens* cpEnum = nullptr;
ULONG ulCount = 0;
hr = SpEnumTokens(SPCAT_VOICES, nullptr, nullptr, &cpEnum);
if (SUCCEEDED(hr)) {
// Get the number of voices.
hr = cpEnum->GetCount(&ulCount);
if (SUCCEEDED (hr))
{
int voiceIndex = var_InheritInteger(p_this, "sapi-voice");
if (voiceIndex > - 1)
{
if (voiceIndex < ulCount) {
hr = cpEnum->Item(voiceIndex, &cpVoiceToken);
if (SUCCEEDED(hr)) {
hr = p_sys->cpVoice->SetVoice(cpVoiceToken);
if (SUCCEEDED(hr)) {
msg_Dbg(p_this, "Selected voice %d", voiceIndex);
}
else {
msg_Err(p_this, "Failed to set voice %d", voiceIndex);
}
cpVoiceToken->Release();
cpVoiceToken = nullptr;
}
}
else
msg_Err(p_this, "Voice index exceeds available count");
}
}
cpEnum->Release();
cpEnum = nullptr;
}
if (SUCCEEDED(hr)) {
hr = p_sys->cpVoice->SetOutput(nullptr, TRUE);
}
}
else
{
msg_Err(p_filter, "Could not create SpVoice");
}
LeaveMTA();
p_filter->pf_render_text = RenderText;
return VLC_SUCCESS;
error:
free(p_sys);
return VLC_EGENERIC;
}
void Sound::test() {
ISpVoice * pVoice = NULL;
ISpObjectToken* pVoiceToken=nullptr;
IEnumSpObjectTokens* pEnum;
ULONG ulCount = 0;
if (FAILED(::CoInitialize(NULL)))
{
return;
}
HRESULT hr = S_OK;
// Find the best matching installed en-us recognizer.
CComPtr<ISpObjectToken> cpRecognizerToken;
if (SUCCEEDED(hr))
{
hr = SpFindBestToken(SPCAT_RECOGNIZERS, L"language=409", NULL, &cpRecognizerToken);
}
// Create the in-process recognizer and immediately set its state to inactive.
CComPtr<ISpRecognizer> cpRecognizer;
if (SUCCEEDED(hr))
{
hr = cpRecognizer.CoCreateInstance(CLSID_SpInprocRecognizer);
}
if (SUCCEEDED(hr))
{
hr = cpRecognizer->SetRecognizer(cpRecognizerToken);
}
if (SUCCEEDED(hr))
{
hr = cpRecognizer->SetRecoState(SPRST_INACTIVE);
}
// Create a new recognition context from the recognizer.
CComPtr<ISpRecoContext> cpContext;
if (SUCCEEDED(hr))
{
hr = cpRecognizer->CreateRecoContext(&cpContext);
}
// Subscribe to the speech recognition event and end stream event.
if (SUCCEEDED(hr))
{
ULONGLONG ullEventInterest = SPFEI(SPEI_RECOGNITION);
hr = cpContext->SetInterest(ullEventInterest, ullEventInterest);
}
// Establish a Win32 event to signal when speech events are available.
HANDLE hSpeechNotifyEvent = INVALID_HANDLE_VALUE;
if (SUCCEEDED(hr))
{
hr = cpContext->SetNotifyWin32Event();
}
if (SUCCEEDED(hr))
{
hSpeechNotifyEvent = cpContext->GetNotifyEventHandle();
if (INVALID_HANDLE_VALUE == hSpeechNotifyEvent)
{
// Notification handle unsupported.
hr = E_NOINTERFACE;
}
}
// Initialize an audio object to use the default audio input of the system and set the recognizer to use it.
CComPtr<ISpAudio> cpAudioIn;
if (SUCCEEDED(hr))
{
hr = cpAudioIn.CoCreateInstance(CLSID_SpMMAudioIn);
}
if (SUCCEEDED(hr))
{
hr = cpRecognizer->SetInput(cpAudioIn, TRUE);
}
// Populate a WAVEFORMATEX struct with our desired output audio format. information.
WAVEFORMATEX* pWfexCoMemRetainedAudioFormat = NULL;
GUID guidRetainedAudioFormat = GUID_NULL;
if (SUCCEEDED(hr))
{
hr = SpConvertStreamFormatEnum(SPSF_16kHz16BitMono, &guidRetainedAudioFormat, &pWfexCoMemRetainedAudioFormat);
}
// Instruct the recognizer to retain the audio from its recognition results.
if (SUCCEEDED(hr))
{
hr = cpContext->SetAudioOptions(SPAO_RETAIN_AUDIO, &guidRetainedAudioFormat, pWfexCoMemRetainedAudioFormat);
}
if (NULL != pWfexCoMemRetainedAudioFormat)
{
CoTaskMemFree(pWfexCoMemRetainedAudioFormat);
}
// Create a new grammar and load an SRGS grammar from file.
CComPtr<ISpRecoGrammar> cpGrammar;
if (SUCCEEDED(hr))
{
hr = cpContext->CreateGrammar(0, &cpGrammar);
}
if (SUCCEEDED(hr))
{
hr = cpGrammar->LoadCmdFromFile(L"grammar.grxml", SPLO_STATIC);
}
// Set all top-level rules in the new grammar to the active state.
if (SUCCEEDED(hr))
{
hr = cpGrammar->SetRuleState(NULL, NULL, SPRS_ACTIVE);
}
// Set the recognizer state to active to begin recognition.
if (SUCCEEDED(hr))
{
hr = cpRecognizer->SetRecoState(SPRST_ACTIVE_ALWAYS);
}
// Establish a separate Win32 event to signal the event loop exit.
HANDLE hExitEvent = CreateEventW(NULL, FALSE, FALSE, NULL);
// Collect the events listened for to pump the speech event loop.
HANDLE rghEvents[] = { hSpeechNotifyEvent, hExitEvent };
// Speech recognition event loop.
BOOL fContinue = TRUE;
while (fContinue && SUCCEEDED(hr))
{
// Wait for either a speech event or an exit event, with a 15 second timeout.
DWORD dwMessage = WaitForMultipleObjects(sp_countof(rghEvents), rghEvents, FALSE, 15000);
switch (dwMessage)
{
// With the WaitForMultipleObjects call above, WAIT_OBJECT_0 is a speech event from hSpeechNotifyEvent.
case WAIT_OBJECT_0:
{
// Sequentially grab the available speech events from the speech event queue.
CSpEvent spevent;
while (S_OK == spevent.GetFrom(cpContext))
{
switch (spevent.eEventId)
{
case SPEI_RECOGNITION:
{
// Retrieve the recognition result and output the text of that result.
ISpRecoResult* pResult = spevent.RecoResult();
LPWSTR pszCoMemResultText = NULL;
hr = pResult->GetText(SP_GETWHOLEPHRASE, SP_GETWHOLEPHRASE, TRUE, &pszCoMemResultText, NULL);
if (SUCCEEDED(hr))
{
wprintf(L"Recognition event received, text=\"%s\"\r\n", pszCoMemResultText);
}
// Also retrieve the retained audio we requested.
CComPtr<ISpStreamFormat> cpRetainedAudio;
if (SUCCEEDED(hr))
{
hr = pResult->GetAudio(0, 0, &cpRetainedAudio);
}
// To demonstrate, we'll speak the retained audio back using ISpVoice.
CComPtr<ISpVoice> cpVoice;
if (SUCCEEDED(hr))
{
hr = cpVoice.CoCreateInstance(CLSID_SpVoice);
}
if (SUCCEEDED(hr))
{
hr = cpVoice->SpeakStream(cpRetainedAudio, SPF_DEFAULT, 0);
}
if (NULL != pszCoMemResultText)
{
CoTaskMemFree(pszCoMemResultText);
}
break;
}
}
}
break;
}
case WAIT_OBJECT_0 + 1:
case WAIT_TIMEOUT:
{
// Exit event or timeout; discontinue the speech loop.
fContinue = FALSE;
//break;
}
}
}
CoUninitialize();
CComPtr <ISpVoice> cpVoice;
CComPtr <ISpStream> cpStream;
CSpStreamFormat cAudioFmt;
//Create a SAPI Voice
hr = cpVoice.CoCreateInstance(CLSID_SpVoice);
//Set the audio format
if (SUCCEEDED(hr))
{
hr = cAudioFmt.AssignFormat(SPSF_22kHz16BitMono);
}
//Call SPBindToFile, a SAPI helper method, to bind the audio stream to the file
if (SUCCEEDED(hr))
{
hr = SPBindToFile(L"c:\\ttstemp.wav", SPFM_CREATE_ALWAYS,
&cpStream, &cAudioFmt.FormatId(), cAudioFmt.WaveFormatExPtr());
}
//set the output to cpStream so that the output audio data will be stored in cpStream
if (SUCCEEDED(hr))
{
hr = cpVoice->SetOutput(cpStream, TRUE);
}
//Speak the text "hello world" synchronously
if (SUCCEEDED(hr))
{
hr = cpVoice->Speak(L"Hello World", SPF_DEFAULT, NULL);
}
//close the stream
if (SUCCEEDED(hr))
{
hr = cpStream->Close();
}
//Release the stream and voice object
cpStream.Release();
cpVoice.Release();
CComPtr<ISpGrammarBuilder> cpGrammarBuilder;
SPSTATEHANDLE hStateTravel;
// Create (if rule does not already exist)
// top-level Rule, defaulting to Active.
hr = cpGrammarBuilder->GetRule(L"Travel", 0, SPRAF_TopLevel | SPRAF_Active, TRUE, &hStateTravel);
// Approach 1: List all possible phrases.
// This is the most intuitive approach, and it does not sacrifice efficiency
// because the grammar builder will merge shared sub-phrases when possible.
// There is only one root state, hStateTravel, and the terminal NULL state,
// and there are six unique transitions between root state and NULL state.
/* XML Approximation:
<rule id="Travel">
<item> fly to Seattle </item>
<item> fly to New York </item>
<item> fly to Washington DC </item>
<item> drive to Seattle </item>
<item> drive to New York </item>
<item> drive to Washington DC </item>
</rule>
*/
// Create set of peer phrases, each containing complete phrase.
// Note: the word delimiter is set as " ", so that the text we
// attach to the transition can be multiple words (for example,
// "fly to Seattle" is implicitly "fly" + "to" + "Seattle"):
if (SUCCEEDED(hr))
{
hr = cpGrammarBuilder->AddWordTransition(hStateTravel, NULL, L"fly to Seattle", L" ", SPWT_LEXICAL, 1, NULL);
}
if (SUCCEEDED(hr))
{
hr = cpGrammarBuilder->AddWordTransition(hStateTravel, NULL, L"fly to New York", L" ", SPWT_LEXICAL, 1, NULL);
}
if (SUCCEEDED(hr))
{
hr = cpGrammarBuilder->AddWordTransition(hStateTravel, NULL, L"fly to Washington DC", L" ", SPWT_LEXICAL, 1, NULL);
}
if (SUCCEEDED(hr))
{
hr = cpGrammarBuilder->AddWordTransition(hStateTravel, NULL, L"drive to Seattle", L" ", SPWT_LEXICAL, 1, NULL);
}
if (SUCCEEDED(hr))
{
hr = cpGrammarBuilder->AddWordTransition(hStateTravel, NULL, L"drive to New York", L" ", SPWT_LEXICAL, 1, NULL);
}
if (SUCCEEDED(hr))
{
hr = cpGrammarBuilder->AddWordTransition(hStateTravel, NULL, L"drive to Washington DC", L" ", SPWT_LEXICAL, 1, NULL);
}
// Find the best matching installed en-US recognizer.
//CComPtr<ISpObjectToken> cpRecognizerToken;
if (SUCCEEDED(hr))
{
hr = SpFindBestToken(SPCAT_RECOGNIZERS, L"language=409", NULL, &cpRecognizerToken);
}
// Create the in-process recognizer and immediately set its state to inactive.
//CComPtr<ISpRecognizer> cpRecognizer;
if (SUCCEEDED(hr))
{
hr = cpRecognizer.CoCreateInstance(CLSID_SpInprocRecognizer);
}
if (SUCCEEDED(hr))
{
hr = cpRecognizer->SetRecognizer(cpRecognizerToken);
}
if (SUCCEEDED(hr))
{
hr = cpRecognizer->SetRecoState(SPRST_INACTIVE);
}
// Create a new recognition context from the recognizer.
//CComPtr<ISpRecoContext> cpContext;
if (SUCCEEDED(hr))
{
hr = cpRecognizer->CreateRecoContext(&cpContext);
}
// Subscribe to the speech recognition event and end stream event.
if (SUCCEEDED(hr))
{
ULONGLONG ullEventInterest = SPFEI(SPEI_RECOGNITION) | SPFEI(SPEI_END_SR_STREAM);
hr = cpContext->SetInterest(ullEventInterest, ullEventInterest);
}
// Establish a Win32 event to signal when speech events are available.
//HANDLE hSpeechNotifyEvent = INVALID_HANDLE_VALUE;
if (SUCCEEDED(hr))
{
hr = cpContext->SetNotifyWin32Event();
}
if (SUCCEEDED(hr))
{
hr = cpContext->SetNotifyWin32Event();
}
if (SUCCEEDED(hr))
{
hSpeechNotifyEvent = cpContext->GetNotifyEventHandle();
if (INVALID_HANDLE_VALUE == hSpeechNotifyEvent)
{
// Notification handle unsupported
//hr = SPERR_UNITIALIZED;
}
}
// Set up an audio input stream using a .wav file and set the recognizer's input.
CComPtr<ISpStream> cpInputStream;
if (SUCCEEDED(hr))
{
hr = SPBindToFile(L"Test.wav", SPFM_OPEN_READONLY, &cpInputStream);
}
if (SUCCEEDED(hr))
{
hr = cpRecognizer->SetInput(cpInputStream, TRUE);
}
// Create a new grammar and load an SRGS grammar from file.
//CComPtr<ISpRecoGrammar> cpGrammar;
if (SUCCEEDED(hr))
{
hr = cpContext->CreateGrammar(0, &cpGrammar);
}
if (SUCCEEDED(hr))
{
hr = cpGrammar->LoadCmdFromFile(L"grammar.grxml", SPLO_STATIC);
}
// Set all top-level rules in the new grammar to the active state.
if (SUCCEEDED(hr))
{
hr = cpGrammar->SetRuleState(NULL, NULL, SPRS_ACTIVE);
}
// Finally, set the recognizer state to active to begin recognition.
if (SUCCEEDED(hr))
{
hr = cpRecognizer->SetRecoState(SPRST_ACTIVE_ALWAYS);
}
hr = CoCreateInstance(CLSID_SpVoice, NULL, CLSCTX_ALL, IID_ISpVoice, (void **)&pVoice);
if (SUCCEEDED(hr)) {
hr = SpEnumTokens(SPCAT_VOICES, L"Gender=Female", NULL, &pEnum);
if (SUCCEEDED(hr))
{
// Get the number of voices.
hr = pEnum->GetCount(&ulCount);
}
// Obtain a list of available voice tokens, set
// the voice to the token, and call Speak.
while (SUCCEEDED(hr) && ulCount--) {
if (pVoiceToken != nullptr) {
pVoiceToken->Release();
}
if (SUCCEEDED(hr))
{
hr = pEnum->Next(1, &pVoiceToken, NULL);
}
if (SUCCEEDED(hr))
{
hr = pVoice->SetVoice(pVoiceToken);
}
if (SUCCEEDED(hr))
{
wchar_t* start = L"<?xml version=\"1.0\" encoding=\"ISO - 8859 - 1\"?><speak version = \"1.0\" xmlns = \"http://www.w3.org/2001/10/synthesis\" xml:lang = \"en-US\">";
wchar_t* end = L"</speak>";
const wchar_t *xml = L"<voice required = \"Gender=Male\"> hi! <prosody pitch=\"fast\"> This is low pitch. </prosody><prosody volume=\"x - loud\"> This is extra loud volume. </prosody>";
wstring s = start;
s += xml;
s += end;
hr = pVoice->Speak(xml, SPF_IS_XML| SPF_ASYNC, 0);
//hr = pVoice->Speak(L"How are you?", SPF_DEFAULT, NULL);
}
}
/*
if (SUCCEEDED(hr)) {
hr = pEnum->Next(1, &pVoiceToken, NULL);
if (SUCCEEDED(hr)) {
hr = pVoice->SetVoice(pVoiceToken);
// Set the output to the default audio device.
if (SUCCEEDED(hr)) {
hr = pVoice->SetOutput(NULL, TRUE);
if (SUCCEEDED(hr)) {
hr = pVoice->Speak(L"Hello, world!", SPF_DEFAULT, 0);
}
}
}
}
*/
pVoice->Release();
}
::CoUninitialize();
}
int _tmain( int argc, _TCHAR* argv[] )
{
cv::setUseOptimized( true );
// Kinectのインスタンス生成、初期化
INuiSensor* pSensor;
HRESULT hResult = S_OK;
hResult = NuiCreateSensorByIndex( 0, &pSensor );
if( FAILED( hResult ) ){
std::cerr << "Error : NuiCreateSensorByIndex" << std::endl;
return -1;
}
hResult = pSensor->NuiInitialize( NUI_INITIALIZE_FLAG_USES_AUDIO );
if( FAILED( hResult ) ){
std::cerr << "Error : NuiInitialize" << std::endl;
return -1;
}
// Audioストリームの初期化(InitializeAudioStream)
std::cout << "InitializeAudioStream" << std::endl;
INuiAudioBeam* pNuiAudioSource;
hResult = pSensor->NuiGetAudioSource( &pNuiAudioSource );
if( FAILED( hResult ) ){
std::cerr << "Error : NuiGetAudioSource" << std::endl;
return -1;
}
IMediaObject* pMediaObject = nullptr;
IPropertyStore* pPropertyStore = nullptr;
pNuiAudioSource->QueryInterface( IID_IMediaObject, reinterpret_cast<void**>( &pMediaObject ) );
pNuiAudioSource->QueryInterface( IID_IPropertyStore, reinterpret_cast<void**>( &pPropertyStore ) );
PROPVARIANT propvariant;
PropVariantInit( &propvariant );
propvariant.vt = VT_I4;
propvariant.lVal = static_cast<LONG>( 4 );
pPropertyStore->SetValue( MFPKEY_WMAAECMA_SYSTEM_MODE, propvariant );
PropVariantClear( &propvariant );
WAVEFORMATEX waveFormat = { AudioFormat, AudioChannels, AudioSamplesPerSecond, AudioAverageBytesPerSecond, AudioBlockAlign, AudioBitsPerSample, 0 };
DMO_MEDIA_TYPE mediaType = { 0 };
MoInitMediaType( &mediaType, sizeof( WAVEFORMATEX ) );
mediaType.majortype = MEDIATYPE_Audio;
mediaType.subtype = MEDIASUBTYPE_PCM;
mediaType.lSampleSize = 0;
mediaType.bFixedSizeSamples = true;
mediaType.bTemporalCompression = false;
mediaType.formattype = FORMAT_WaveFormatEx;
memcpy( mediaType.pbFormat, &waveFormat, sizeof( WAVEFORMATEX ) );
pMediaObject->SetOutputType( 0, &mediaType, 0 );
KinectAudioStream* audioStream = new KinectAudioStream( pMediaObject );
IStream* pStream = nullptr;
audioStream->QueryInterface( IID_IStream, reinterpret_cast<void**>( &pStream ) );
CoInitialize( nullptr );
ISpStream* pSpeechStream = nullptr;
CoCreateInstance( CLSID_SpStream, NULL, CLSCTX_INPROC_SERVER, __uuidof(ISpStream), reinterpret_cast<void**>( &pSpeechStream ) );
pSpeechStream->SetBaseStream( pStream, SPDFID_WaveFormatEx, &waveFormat );
MoFreeMediaType( &mediaType );
pStream->Release();
pPropertyStore->Release();
pMediaObject->Release();
pNuiAudioSource->Release();
// 音声認識器を作成(CreateSpeechRecognizer)
std::cout << "CreateSpeechRecognizer" << std::endl;
ISpRecognizer* pSpeechRecognizer;
CoCreateInstance( CLSID_SpInprocRecognizer, nullptr, CLSCTX_INPROC_SERVER, __uuidof(ISpRecognizer), reinterpret_cast<void**>( &pSpeechRecognizer ) );
pSpeechRecognizer->SetInput( pSpeechStream, false );
/*
// If can use ATL, easier to using SpFindBestToken(sphelper.h). When using Professional or more.
ISpObjectToken* pEngineToken = nullptr;
SpFindBestToken( SPCAT_RECOGNIZERS, L"Language=411;Kinect=True", NULL, &pEngineToken ); // Japanese "Language=411;Kinect=True" English "Language=409;Kinect=True"
*/
///*
// If can't use ATL, alternative to using SpFIndBestToken(sphelper.h). When using Express.
const wchar_t* pVendorPreferred = L"VendorPreferred";
const unsigned long lengthVendorPreferred = static_cast<unsigned long>( wcslen( pVendorPreferred ) );
unsigned long length;
ULongAdd( lengthVendorPreferred, 1, &length );
wchar_t* pAttribsVendorPreferred = new wchar_t[ length ];
StringCchCopyW( pAttribsVendorPreferred, length, pVendorPreferred );
ISpObjectTokenCategory* pTokenCategory = nullptr;
CoCreateInstance( CLSID_SpObjectTokenCategory, nullptr, CLSCTX_ALL, __uuidof(ISpObjectTokenCategory), reinterpret_cast<void**>( &pTokenCategory ) );
pTokenCategory->SetId( SPCAT_RECOGNIZERS, false );
IEnumSpObjectTokens* pEnumTokens = nullptr;
CoCreateInstance( CLSID_SpMMAudioEnum, nullptr, CLSCTX_ALL, __uuidof(IEnumSpObjectTokens), reinterpret_cast<void**>( &pEnumTokens ) );
pTokenCategory->EnumTokens( L"Language=411;Kinect=True", pAttribsVendorPreferred, &pEnumTokens ); // Japanese "Language=411;Kinect=True" English "Language=409;Kinect=True"
delete[] pAttribsVendorPreferred;
ISpObjectToken* pEngineToken = nullptr;
pEnumTokens->Next( 1, &pEngineToken, nullptr );
//*/
pSpeechRecognizer->SetRecognizer( pEngineToken );
ISpRecoContext* pSpeechContext;
pSpeechRecognizer->CreateRecoContext( &pSpeechContext );
pEngineToken->Release();
///*
pTokenCategory->Release();
pEnumTokens->Release();
//*/
// 音声認識辞書の作成(LoadSpeechGrammar)
std::cout << "LoadSpeechGrammar" << std::endl;
ISpRecoGrammar* pSpeechGrammar;
pSpeechContext->CreateGrammar( 1, &pSpeechGrammar );
pSpeechGrammar->LoadCmdFromFile( L"SpeechRecognition_Ja.grxml", /*SPLO_STATIC*/SPLO_DYNAMIC ); // http://www.w3.org/TR/speech-grammar/ (UTF-8/CRLF)
audioStream->StartCapture();
pSpeechGrammar->SetRuleState( nullptr, nullptr, SPRS_ACTIVE );
pSpeechRecognizer->SetRecoState( SPRST_ACTIVE_ALWAYS );
pSpeechContext->SetInterest( SPFEI( SPEI_RECOGNITION ), SPFEI( SPEI_RECOGNITION ) );
pSpeechContext->Resume( 0 );
HANDLE hSpeechEvent = INVALID_HANDLE_VALUE;
hSpeechEvent = pSpeechContext->GetNotifyEventHandle();
HANDLE hEvents[1] = { hSpeechEvent };
int width = 640;
int height = 480;
cv::Mat audioMat = cv::Mat::zeros( height, width, CV_8UC3 );
cv::namedWindow( "Audio" );
bool exit = false;
std::cout << std::endl << "Speech Recognition Start..." << std::endl << std::endl;
while( 1 ){
// イベントの更新待ち
ResetEvent( hSpeechEvent );
unsigned long waitObject = MsgWaitForMultipleObjectsEx( ARRAYSIZE( hEvents ), hEvents, INFINITE, QS_ALLINPUT, MWMO_INPUTAVAILABLE );
if( waitObject == WAIT_OBJECT_0 ){
// イベントの取得
const float confidenceThreshold = 0.3f;
SPEVENT eventStatus;
unsigned long eventFetch = 0;
pSpeechContext->GetEvents( 1, &eventStatus, &eventFetch );
while( eventFetch > 0 ){
switch( eventStatus.eEventId ){
// 音声認識イベント(SPEI_HYPOTHESIS:推定またはSPEI_RECOGNITION:認識)
case SPEI_HYPOTHESIS:
case SPEI_RECOGNITION:
if( eventStatus.elParamType == SPET_LPARAM_IS_OBJECT ){
// フレーズの取得
ISpRecoResult* pRecoResult = reinterpret_cast<ISpRecoResult*>( eventStatus.lParam );
SPPHRASE* pPhrase = nullptr;
hResult = pRecoResult->GetPhrase( &pPhrase );
if( SUCCEEDED( hResult ) ){
if( ( pPhrase->pProperties != nullptr ) && ( pPhrase->pProperties->pFirstChild != nullptr ) ){
// 辞書のフレーズタグと比較
const SPPHRASEPROPERTY* pSemantic = pPhrase->pProperties->pFirstChild;
if( pSemantic->SREngineConfidence > confidenceThreshold ){
if( wcscmp( L"あか", pSemantic->pszValue ) == 0 ){
std::cout << "あか" << std::endl;
audioMat = cv::Scalar( 0, 0, 255 );
}
else if( wcscmp( L"みどり", pSemantic->pszValue ) == 0 ){
std::cout << "みどり" << std::endl;
audioMat = cv::Scalar( 0, 255, 0 );
}
else if( wcscmp( L"あお", pSemantic->pszValue ) == 0 ){
std::cout << "あお" << std::endl;
audioMat = cv::Scalar( 255, 0, 0 );
}
else if( wcscmp( L"おわり", pSemantic->pszValue ) == 0 ){
exit = true;
}
}
}
CoTaskMemFree( pPhrase );
}
}
break;
default:
break;
}
pSpeechContext->GetEvents( 1, &eventStatus, &eventFetch );
}
}
// 表示
cv::imshow( "Audio", audioMat );
// ループの終了判定(Escキー)
if( cv::waitKey( 30 ) == VK_ESCAPE || exit ){
break;
}
}
// 終了処理
audioStream->StopCapture();
pSpeechRecognizer->SetRecoState( SPRST_INACTIVE );
CoUninitialize();
pSensor->NuiShutdown();
CloseHandle( hSpeechEvent );
cv::destroyAllWindows();
return 0;
}
