This documentation is archived and is not being maintained.

SpeechRecognitionEngine.SpeechRecognized Event

The event raised when the recognition engine detects speech, and has found one or more phrases with sufficient confidence levels.

Namespace: Microsoft.Speech.Recognition
Assembly: Microsoft.Speech (in microsoft.speech.dll)

The example below shows the subscription of an anonymous method as a handler for SpeechRecognized, which in turn calls a display function (DisplayResult) to return to the user interface the best candidate and the list of alternates generated.

A second anonymous method subscribes to the SpeechRecognized event of the colorChooserGrammar instance of Grammar.

If the colorChooserGrammar Grammar is used to recognize the best candidate phrase, its handler for SpeechRecognized will be called prior to the handler of the SpeechRecognized eveny for the SpeechRecognitionEngine object.

_recognizer.SpeechRecognized +=
delegate(object sender, SpeechRecognizedEventArgs eventArgs) {
    if (!_asyncEmulation) {

Utils.DisplayAudioInputFormat(_audioStateLabel, _recognizer);
DisplayResult(eventArgs);
    }
};

colorChooserGrammar.SpeechRecognized +=
    delegate(object sender, SpeechRecognizedEventArgs eventArgs) {
// Retrieve the value of the semantic property.
string colorName = (string)eventArgs.Result.Semantics["color"].Value;
BackColor = Color.FromName(colorName);
float Bright = BackColor.GetBrightness();
float Hue = BackColor.GetHue();
float Sat = BackColor.GetSaturation();
//Make sure the text is readable regardless of background.
if (BackColor.GetBrightness() <= .50) {
    ForeColor = Color.White;
} else {
    ForeColor = Color.Black;
}
        };
// Displays the semantics, rules, and alternates of the result.
private void DisplayResult(EventArgs eventArgs) {

    //Make sure We handle different event types we support
     RecognitionResult result=null;
    if (null !=(eventArgs as RecognitionEventArgs)) {
result = (eventArgs as RecognitionEventArgs).Result;
    }else if (null !=(eventArgs as EmulateRecognizeCompletedEventArgs)) {
result = (eventArgs as EmulateRecognizeCompletedEventArgs).Result;
    }

    // In all cases, clear results, semantic XML, semantics tree, and alternates.
    _recognitionSmlBrowser.Navigate("about:blank");
    _semanticsTreeView.Nodes.Clear();
    //            _candidateList.Items.Clear();
    _candidateList.DataSource = null;
    _candidateList.DisplayMember = null;
    _phraseInfoLabel.Text = "";
    _bestCandidateTextBox.Text = "";

    if (result != null) {
//If eventArgs not null repopulate displays
//Obtain result from RecognitionEventArg passed in.


//Set Text color on the basis of the event type passed in.

switch (eventArgs.GetType().ToString()) {

    case "Microsoft.Speech.Recognition.SpeechHypothesizedEventArgs":
        _bestCandidateTextBox.ForeColor = Color.Black;
        _bestCandidateTextBox.Text = result.Text;
        _phraseInfoLabel.Text = null;
        break;
    case "Microsoft.Speech.Recognition.SpeechRecognitionRejectedEventArgs":
        _bestCandidateTextBox.ForeColor = Color.OrangeRed;
        _bestCandidateTextBox.Text = "Unable to Recognize Input";
        _phraseInfoLabel.Text = String.Format("Rejected Phrase Information\n");
        Utils.DisplayBasicPhraseInfo(_phraseInfoLabel, result);
        Utils.DisplaySemanticsSML(_recognitionSmlBrowser, result);
        Utils.DisplaySemanticsTreeView(_semanticsTreeView, result.Semantics);
        break;
    case "Microsoft.Speech.Recognition.SpeechRecognizedEventArgs":
        _bestCandidateTextBox.ForeColor = Color.Green;
        _bestCandidateTextBox.Text = result.Text;
        _phraseInfoLabel.Text = String.Format("Recognized Phrase Information\n");
        Utils.DisplayBasicPhraseInfo(_phraseInfoLabel, result);
        Utils.DisplaySemanticsSML(_recognitionSmlBrowser, result);
        Utils.DisplaySemanticsTreeView(_semanticsTreeView, result.Semantics);
        break;
    case "Microsoft.Speech.Recognition.EmulateRecognizeCompletedEventArgs":
        _bestCandidateTextBox.ForeColor = Color.Green;
        _bestCandidateTextBox.Text = result.Text;
        _phraseInfoLabel.Text = String.Format("Recognized Phrase Information\n");
        Utils.DisplayBasicPhraseInfo(_phraseInfoLabel, result);
        Utils.DisplaySemanticsSML(_recognitionSmlBrowser, result);
        Utils.DisplaySemanticsTreeView(_semanticsTreeView, result.Semantics);
        break;
    default:
        _bestCandidateTextBox.ForeColor = Color.Black;
        break;
}

Utils.DisplayDataOnListBox(_candidateList, result.Alternates, "Text");

    }
}

The Grammar class also supports a SpeechRecognized event (SpeechRecognized).

Handlers subscribed to the SpeechRecognized event of the instance of Grammar used to recognize a given candidate phrase are always called prior calling the handlers for SpeechRecognized. Any tasks specific to a particular grammar should always be handled by SpeechRecognized event handlers.

Handlers of SpeechRecognized events can obtain the best possible candidate phrase, as well as a list of up to MaxAlternates phrases with lower confidence levels through an instance of the RecognitionResult object returned by the Result property of the SpeechRecognizedEventArgs object passed to the handler. (SpeechRecognitionRejectedEventArgs is derived fromSpeechRecognitionEventArgs)).

The confidence level cut off for recognition is an arbitrary feature of any particular recognition engine and is not settable by applications.


Development Platforms

Windows XP Professional with Service Pack 2 (SP2), Windows Server 2003, Windows Vista Ultimate Edition, Windows Vista Business Edition, Windows Vista Enterprise Edition

Target Platforms

Show: