摘選自：“北極悠藍”的博客《C++使用SAPI實現語音合成和語音識別的方法和代碼》
微軟的語音識別,在這里我們簡稱它為SR(speech recognition),SR分為兩種模式的監聽：第一種模式：聽寫模式，即隨意輸入語音，監聽對象將最為接近的字或者詞，句反饋出來；第二種模式：命令與控制模式，劃定范圍監聽，制定一組被選項做為監聽的，用戶的語音輸入被反饋成最為接近的一個選項。說得通俗一些：第一種是填空題，第二種是選擇題目。
下面是第一種模式的代碼：
#include <windows.h>
#include <sapi.h>
#include <stdio.h>
#include <string.h>
#include <atlbase.h>
#include "sphelper.h" inline HRESULT BlockForResult(ISpRecoContext * pRecoCtxt, ISpRecoResult ** ppResult)
{HRESULT hr = S_OK;CSpEvent event;
while (SUCCEEDED(hr) && SUCCEEDED(hr = event
.GetFrom(pRecoCtxt)) && hr == S_FALSE){hr = pRecoCtxt->WaitForNotifyEvent(INFINITE);}*ppResult = event
.RecoResult();
if (*ppResult){(*ppResult)->AddRef();}
return hr;
}
const WCHAR * StopWord()
{
const WCHAR * pchStop;LANGID LangId = ::SpGetUserDefaultUILanguage();
switch (LangId){
case MAKELANGID(LANG_JAPANESE, SUBLANG_DEFAULT):pchStop = L
"}42N86\0b70e50fc0ea0e70fc/05708504608a087046";;
break;
default:pchStop = L
"Stop";
break;}
return pchStop;
}
int main(
int argc,
char* argv[])
{HRESULT hr = E_FAIL;
bool fUseTTS =
true;
bool fReplay =
true;
if (argc >
1){
int i;
for (i =
1; i < argc; i++){
if (_stricmp(argv[i],
"-noTTS") ==
0){fUseTTS =
false;
continue;}
if (_stricmp(argv[i],
"-noReplay") ==
0){fReplay =
false;
continue;}printf(
"Usage: %s [-noTTS] [-noReplay] ", argv[
0]);
return hr;}}
if (SUCCEEDED(hr = ::CoInitialize(
NULL))){{CComPtr<ISpRecoContext> cpRecoCtxt;CComPtr<ISpRecoGrammar> cpGrammar;CComPtr<ISpVoice> cpVoice;hr = cpRecoCtxt
.CoCreateInstance(CLSID_SpSharedRecoContext);
if (SUCCEEDED(hr)){hr = cpRecoCtxt->GetVoice(&cpVoice);}
if (cpRecoCtxt && cpVoice &&SUCCEEDED(hr = cpRecoCtxt->SetNotifyWin32Event()) &&SUCCEEDED(hr = cpRecoCtxt->SetInterest(SPFEI(SPEI_RECOGNITION), SPFEI(SPEI_RECOGNITION))) &&SUCCEEDED(hr = cpRecoCtxt->SetAudioOptions(SPAO_RETAIN_AUDIO,
NULL,
NULL)) &&SUCCEEDED(hr = cpRecoCtxt->CreateGrammar(
0, &cpGrammar)) &&SUCCEEDED(hr = cpGrammar->LoadDictation(
NULL, SPLO_STATIC)) &&SUCCEEDED(hr = cpGrammar->SetDictationState(SPRS_ACTIVE))){USES_CONVERSION;
const WCHAR *
const pchStop = StopWord();CComPtr<ISpRecoResult> cpResult;printf(
"I will repeat everything you say. Say %s to exit. ", W2A(pchStop));
while (SUCCEEDED(hr = BlockForResult(cpRecoCtxt, &cpResult))){cpGrammar->SetDictationState(SPRS_INACTIVE);CSpDynamicString dstrText;
if (SUCCEEDED(cpResult->GetText(SP_GETWHOLEPHRASE, SP_GETWHOLEPHRASE,
TRUE, &dstrText,
NULL))){printf(
"I heard: %s ", W2A(dstrText));
if (fUseTTS){cpVoice->Speak(L
"I heard", SPF_ASYNC,
NULL);cpVoice->Speak(dstrText, SPF_ASYNC,
NULL);}
if (fReplay){
if (fUseTTS)cpVoice->Speak(L
"when you said", SPF_ASYNC,
NULL);
elseprintf(
" when you said ");cpResult->SpeakAudio(
NULL,
0,
NULL,
NULL);}cpResult
.Release();}
if (_wcsicmp(dstrText, pchStop) ==
0){
break;}cpGrammar->SetDictationState(SPRS_ACTIVE);}}}::CoUninitialize();}
return hr;
}
總結
以上是生活随笔為你收集整理的用SAPI实现Speech Recognition（SR） - 听写模式的全部內容，希望文章能夠幫你解決所遇到的問題。
如果覺得生活随笔網站內容還不錯，歡迎將生活随笔推薦給好友。
