How can I change the code to English text to speak instead of Japanese?

At the moment, I have a code for text to speech, but I want to change the code so the output will be in English instead of a Japanese speaker what is currently the case. Is anyone familiar with this and can help me

// SAPIの初期化
if (FAILED(CoCreateInstance(CLSID_SpVoice, NULL, CLSCTX_ALL, IID_ISpVoice, (void **)&pVoice))) 
{
    throw;
}


// Register events to detect
pVoice->SetInterest(SPFEI(SPEI_PHONEME) | SPFEI(SPEI_END_INPUT_STREAM),
    SPFEI(SPEI_PHONEME) | SPFEI(SPEI_END_INPUT_STREAM));
    pVoice->SetNotifyWindowMessage(hWnd, WM_USER, 0, 0);    // subscribe message

// preprocessing to speak
pVoice->Speak( NULL , SPF_PURGEBEFORESPEAK , 0 );   

}

XML mode

hr = voice->Speak(
        L"<sapi><voice required=\"Language=409\">Hello World</voice></sapi>",
        SPF_IS_XML,
        0 );

or

hr = voice->Speak(
        L"<sapi><lang langid=\"409\">Hello World</lang></sapi>",
        SPF_IS_XML,
        0 );

Handcoded mode

bool SetFirstFitNarrator( ISpVoice* voice, const WCHAR* language )
{
    HRESULT                         hr;
    ISpObjectToken*                 token;
    CComPtr<IEnumSpObjectTokens>    cpEnum;

    hr = SpEnumTokens( SPCAT_VOICES, language, 0, &cpEnum );
    if ( hr != S_OK )
        return false;

    while ( cpEnum->Next( 1, &token, NULL ) == S_OK )
    {
        // We have narrator handling this language, try it
        HRESULT hr = voice->SetVoice( token );
        if ( hr == S_OK )
        {
            // Some sanity tests, in most cases not nessesary.
            // I'm using them to detect broken narrators.
            // You can just
            //    return true;
            // instead.
            hr = voice->Speak( 0, SPF_PURGEBEFORESPEAK, 0 );
            if ( hr == S_OK )
            {
                hr = voice->Speak( L" ", SPF_ASYNC | SPF_IS_NOT_XML, 0 );
                if ( hr == S_OK )
                {
                    // Narrator seems to work OK
                    retrun true;
                }
            }
        }

        // Alternatively you can collect narrators in some table
        // and offer them to user for selection.
        //
        // This is how you can obtain user friendly description, eg.
        // "Microsoft David Desktop - English (United States)"
        //
        //  CSpDynamicString dstrDesc;
        //  hr = SpGetDescription( token, &dstrDesc );
        //  if ( hr == S_OK )
        //  {
        //      const WCHAR*    user_friendly_description = dstrDesc;
    }

    // None of the narrators can handle given language
    return false;
}

void Test()
{
    // "Language=409" - means US English
    if ( SetFirstFitNarrator( voice, L"Language=409" ) )
        voice->Speak( L"hello world ", SPF_IS_NOT_XML, 0 );
    else
    {
        MessageBox(
                hwnd,
                L"Unfortunately there is no narrator handling US English language.",
                L"No narrator",
                MB_OK | MB_ICONEXCLAMATION );
    }
}

For language ids download PDF from [MS-LCID]: Windows Language Code Identifier (LCID) Reference