tts 语音库开发

本文转摘自网络，写得不错。

 

Hello World by Microsoft Speech SDK 5.1

 
Speech SDK 采用 COM 的形式封装，和大多数COM对象一样，下面是一个似曾相识的初始化：
 
    CoInitialize(NULL);
 
    CLSIDFromProgID(L"SAPI.SpVoice", &CLSID_SpVoice);
 
    ISpVoice * pSpVoice = NULL;
 
    if (FAILED(CoCreateInstance(CLSID_SpVoice, NULL,
        CLSCTX_INPROC_SERVER, IID_ISpVoice, (void **)&pSpVoice)))
    {
        cout << "Failed to create instance of ISpVoice!" << endl;
        return -1;
    }
 
    现在我们用它来说 “Hello World”，非常简单：
 
pSpVoice->Speak(L"Hello World!", SPF_DEFAULT, NULL);
 
最后，清除这一切
 
pSpVoice->Release();
 
CoUninitialize();
 
很容易，是不是？下面我们稍微改变一下：
 
IEnumSpObjectTokens *pSpEnumTokens = NULL;
if (SUCCEEDED(SpEnumTokens(SPCAT_VOICES, L"language=409", NULL, &pSpEnumTokens)))
{
       ISpObjectToken *pSpToken = NULL;
 
       while (SUCCEEDED(pSpEnumTokens->Next(1, &pSpToken, NULL)) &&
              pSpToken != NULL)
       {
              pSpVoice->SetVoice(pSpToken);
              pSpVoice->Speak(L"Hello World!", SPF_DEFAULT, NULL);
 
              pSpToken->Release();
       }
 
       pSpEnumTokens->Release();
}
 
如果只需要一种 Voice，SpFindBestToken 更加方便：
 
ISpObjectToken * pSpObjectToken = NULL;
 
if (SUCCEEDED(SpFindBestToken(SPCAT_VOICES, L"gender=female",
       NULL, &pSpObjectToken)))
{
       pSpVoice->SetVoice(pSpObjectToken);
 
       pSpVoice->Speak(L"Hello World!", SPF_DEFAULT, NULL);
 
       pSpObjectToken->Release();
}
 
现在改为使用中文：
 
ISpObjectToken * pSpObjectToken = NULL;
 
if (SUCCEEDED(SpFindBestToken(SPCAT_VOICES, L"language=804",
       NULL, &pSpObjectToken)))
{
       pSpVoice->SetVoice(pSpObjectToken);
 
       pSpVoice->Speak(L"世界你好！", SPF_DEFAULT, NULL);
 
       pSpObjectToken->Release();
}
 
ISpVoice::Speak 可以允许朗读的句子中包含 XML 标记，其实以上这一切用 XML 标记将变得十分简单：
 
pSpVoice->Speak(L"世界你好！", SPF_DEFAULT, NULL);
 
    pSpVoice->Speak(L"Hello World!", SPF_DEFAULT, NULL);
 
最后，以一段简单的报时结束这篇文章：
 
SYSTEMTIME st;
GetLocalTime(&st);
 
ISpObjectToken * pSpObjectToken = NULL;
if (SUCCEEDED(SpFindBestToken(SPCAT_VOICES, L"language=804", NULL, &pSpObjectToken)))
{
       pSpVoice->SetVoice(pSpObjectToken);
 
       WCHAR wsz[128];
       swprintf(wsz,
              L"现在是%.4d-%.02d-%.2d %.2d:%.02d",
              st.wYear, st.wMonth, st.wDay, st.wHour, st.wMinute);
 
       pSpVoice->Speak(wsz, SPF_IS_XML, NULL);
 
       pSpObjectToken->Release();
}
http://blog.csdn.net/lkjx82/archive/2006/07/23/965120.aspx

现在语音技术应用十分广泛，本人也一直在从事这方面的工作，利用MS TTS开发产品，针对中文发音设置的问题，在这里阐述一下自己的实现方法。

void CIMEInfoAcquisitionDlg::SpeakVoice(LPCTSTR csVoice)
{
   //com初期化
    if (FAILED(CoInitialize(NULL)))
    {
        AfxMessageBox("Error to intiliaze COM");
        return;
    }
 
 //文字変換
 LPOLESTR lpw = new WCHAR [256];
    MultiByteToWideChar(CP_ACP,0,csVoice,-1,lpw,256);

 HRESULT hr = S_OK;
    if( SUCCEEDED( hr ) )
 {
  CComPtr        cpVoiceToken;
  CComPtr   cpEnum;
  CComPtr              cpVoice;
  ULONG                          ulCount = 0;

  // 建立

  hr = cpVoice.CoCreateInstance(CLSID_SpVoice);

  LPCSTR name;
    name = "Microsoft Simplified Chinese";

  CString s; s.Format("Name=%s", name);
  WCHAR wName[1024];
  MultiByteToWideChar(CP_ACP, 0, s, -1, wName, sizeof(wName));

  ISpObjectToken* pToken = NULL;
  if(SUCCEEDED(SpFindBestToken(SPCAT_VOICES, NULL, wName, &pToken))){
  cpVoice->SetVoice(pToken);
  }
  //文读
     if (SUCCEEDED (hr))
  {
     hr = cpVoice->Speak( lpw, SPF_DEFAULT, NULL );
  }

 }

 //com的结束
    delete [] lpw;
    CoUninitialize();
 return;
}

以上利用ISpObjectToken接口实现了对中文的阅读，是不是很简单明了：）

http://blog.csdn.net/lkjx82/archive/2006/07/23/965120.aspx

 

本文来自CSDN博客，转载请标明出处：http://blog.csdn.net/chrisyue/archive/2007/11/19/1893166.aspx