Recognizing Speech in Other Languages

The market for Windows Phone apps extends across many countries and regions. Supporting other languages is sometimes critical to effectively monetizing your app. Fortunately, the speech features of the Windows Phone SDK make it possible for you to provide speech capabilities for numerous languages.

In this section we look at selecting a language and region-specific speech recognizer to recognize speech in a non-English language.

The sample for this section is located in the Speech/NonEnglishRecognizer directory of the WPUnleashed.Examples project in the downloadable sample code.

Windows Phone 8 supports speech recognition and Text-to-Speech (TTS) for the languages and regions in Table 23.1.

TABLE 23.1. Speech Recognition Languages

A default speech recognizer language is selected automatically by the OS, depending on the language settings of the device. You can, however, substitute a different language by calling SetRecognizer on the SpeechRecognizer instance. A language-specific SpeechRecognizerInformation can be retrieved from the static InstalledSpeechRecognizers.All property, as shown in Listing 23.12.

LISTING 23.12. NonEnglishViewModel.GetSpeechRecognizerUI Method

Click here to view code image

SpeechRecognizerUI GetSpeechRecognizerUI()
{
    if (recognizerUI == null)
    {
        recognizerUI = new SpeechRecognizerUI();
        recognizerUI.Settings.ReadoutEnabled = true;

        IEnumerable<SpeechRecognizerInformation> frenchRecognizers
            = from recognizerInfo in InstalledSpeechRecognizers.All
                where recognizerInfo.Language == "fr-FR"
                select recognizerInfo;

        recognizerUI.Recognizer.SetRecognizer(frenchRecognizers.Single());


        /* Create a list grammar from the string array
         * and add it to the grammar set. */
        recognizerUI.Recognizer.Grammars.AddGrammarFromList(
                                             "frenchNumbers", frenchNumbers);
    }

    return recognizerUI;
}

In this example, a list grammar is used to define the set of words that the speech recognizer listens for. The viewmodel contains the following array of numbers in French:

Click here to view code image

readonly string[] frenchNumbers = { "un", "deux", "trois", "quatre", "cinq",
"six", "sept", "huit", "neuf", "dix" };

The use of a speech recognizer that has been initialized for a nondefault language remains the same. The ListenText and ExampleText properties of the recognizer’s Settings should be set to strings in the applicable language. See Listing 23.13.

LISTING 23.13. NonEnglishViewModel.Prompt Method

Click here to view code image

async void Prompt()
{
    SpeechRecognizerUI recognizer = GetSpeechRecognizerUI();

    recognizer.Settings.ListenText = "Say a French number.";
    recognizer.Settings.ExampleText = " 'un', 'deux', 'trois', 'quatre' ";

    SpeechRecognitionUIResult uiResult = await recognizer.RecognizeWithUIAsync();

    if (uiResult.ResultStatus == SpeechRecognitionUIStatus.Succeeded)
    {
        RecognizedText = uiResult.RecognitionResult.Text;
    }
}

There is no limit placed on combining different languages in your app, which allows you to build rich, multilingual applications.