Windows TTS无法切换男性语音包问题求助

问题重现 #

尝试通过gender=male筛选语音播放音频失败，手动添加了Microsoft Kangkang（中文男声）的注册表项：

Windows Registry Editor Version 5.00

[HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Speech\Voices\Tokens\TTS_MS_ZH-CN_KANGKANG_11.0]
@="Microsoft Kangkang - Chinese (Simplified, PRC)"
"804"="Microsoft Kangkang - Chinese (Simplified, PRC)"
"CLSID"="{179F3D56-1B0B-42B2-A962-59B7EF59FE1B}"
"LangDataPath"=hex(2):25,00,77,00,69,00,6e,00,64,00,69,00,72,00,25,00,5c,00,53,\
  00,70,00,65,00,65,00,63,00,68,00,5f,00,4f,00,6e,00,65,00,43,00,6f,00,72,00,\
  65,00,5c,00,45,00,6e,00,67,00,69,00,6e,00,65,00,73,00,5c,00,54,00,54,00,53,\
  00,5c,00,7a,00,68,00,2d,00,43,00,4e,00,5c,00,4d,00,53,00,54,00,54,00,53,00,\
  4c,00,6f,00,63,00,7a,00,68,00,43,00,4e,00,2e,00,64,00,61,00,74,00,00,00
"VoicePath"=hex(2):25,00,77,00,69,00,6e,00,64,00,69,00,72,00,25,00,5c,00,53,00,\
  70,00,65,00,65,00,63,00,68,00,5f,00,4f,00,6e,00,65,00,43,00,6f,00,72,00,65,\
  00,5c,00,45,00,6e,00,67,00,69,00,6e,00,65,00,73,00,5c,00,54,00,54,00,53,00,\
  5c,00,7a,00,68,00,2d,00,43,00,4e,00,5c,00,4d,00,32,00,30,00,35,00,32,00,59,\
  00,61,00,6f,00,79,00,61,00,6f,00,00,00

[HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Speech\Voices\Tokens\TTS_MS_ZH-CN_KANGKANG_11.0\Attributes]
"Age"="Adult"
"DataVersion"="11.0.2013.1022"
"Gender"="Male"
"Language"="804"
"Name"="Microsoft Kangkang"
"SayAsSupport"="spell=NativeSupported; cardinal=GlobalSupported; ordinal=NativeSupported; date=GlobalSupported; time=GlobalSupported; telephone=NativeSupported; computer=NativeSupported; address=NativeSupported; percentage=NativeSupported; currency=NativeSupported; message=NativeSupported; url=NativeSupported; alphanumeric=NativeSupported"
"SharedPronunciation"=""
"Vendor"="Microsoft"
"Version"="11.0"

导入注册表后，通过以下代码枚举语音：

ULONG len;
m_pIEnumSpObjectTokens->GetCount(&len);
for (auto i = 0; i < len; ++i) {
    ISpObjectToken* token;
    m_pIEnumSpObjectTokens->Item(i, &token);
    WCHAR* name;
    token->GetId(&name);
    OutputDebugStringW(name);
}

输出可见Kangkang已被识别：

HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Speech\Voices\Tokens\TTS_MS_ZH-CN_HUIHUI_11.0
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Speech\Voices\Tokens\TTS_MS_EN-US_ZIRA_11.0
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Speech\Voices\Tokens\TTS_MS_ZH-CN_KANGKANG_11.0

但使用以下代码设置男声后，仍播放默认语音：

if (iGender == 1){
    if (SUCCEEDED(SpFindBestToken(SPCAT_VOICES, L"gender=male", NULL, &pSpToken)))
        pVoice->SetVoice(pSpToken);
}

解决方案 #

1. 检查语音文件是否存在
   注册表中VoicePath指向的文件%windir%\Speech_OneCore\Engines\TTS\zh-CN\M2052Yayaao需要实际存在。如果文件缺失，即使注册表项正确，语音也无法加载，会自动 fallback 到默认语音。

2. 明确指定语言筛选条件
   SpFindBestToken可能优先匹配与当前系统语言或语音合成器默认语言一致的语音。添加语言筛选器，确保找到中文男声：

   if (iGender == 1){
       // 添加语言筛选（804对应中文简体）
       if (SUCCEEDED(SpFindBestToken(SPCAT_VOICES, L"gender=male;language=804", NULL, &pSpToken)))
           pVoice->SetVoice(pSpToken);
   }
   

3. 验证SetVoice调用是否成功
   不要只检查SpFindBestToken的返回值，也要确认SetVoice是否成功：

   if (iGender == 1){
       if (SUCCEEDED(SpFindBestToken(SPCAT_VOICES, L"gender=male;language=804", NULL, &pSpToken)))
       {
           HRESULT hr = pVoice->SetVoice(pSpToken);
           if (FAILED(hr))
           {
               // 输出错误码，排查问题
               wchar_t errMsg[256];
               swprintf_s(errMsg, L"SetVoice failed: 0x%08X", hr);
               OutputDebugStringW(errMsg);
           }
       }
   }
   

4. 重启语音合成服务或应用
   导入注册表后，部分情况下需要重启应用或Windows语音服务（SpeechSynthesizer相关服务）才能让新的语音项生效。

5. 检查语音位数兼容性
   如果你的应用是32位，需要确保注册表项也写入了HKEY_LOCAL_MACHINE\SOFTWARE\Wow6432Node\Microsoft\Speech\Voices\Tokens路径（64位系统下32位程序会读取这个路径的注册表）。可以把相同的注册表项复制到该路径下。

内容的提问来源于stack exchange，提问作者Ice Wind