Bruno Silva

I’ve been wanting to try out System.Speech namespace (present in 3.0 and 3.5 .NET Framework) but unfortunately until this weekend I was stuck with a Portuguese version of Windows Vista which prevented me from using Speech features.

Today I finally tried it out and I was stunned! It couldn’t be simpler (at least for a simple example ). I’ll explain in a few lines of code how to use two main features: Speech Synthesis and Speech Recognition.

Speech Synthesis

using System.Speech.Synthesis;

To make your computer to say something is as simple as 2 lines of code can be:

synthesizer = new SpeechSynthesizer();

synthesizer.Speak("Hello World");

This piece of code has one disadvantage: it locks your program until “Hello World” is said. You can achieve the same goal without this problem by using a asynchronous approach.

// Shut up any speaking going on

if (synthesizer.State == SynthesizerState.Speaking)

    synthesizer.SpeakAsyncCancel(

        synthesizer.GetCurrentlySpokenPrompt());

// Say text (Async to prevent blocking)

if (synthesizer.State == SynthesizerState.Ready)

        synthesizer.SpeakAsync("Hello World");

Speech Recognition

using System.Speech.Recognition;

Speech recognition requires a setup with more code, but as simples as synthesis.

 sre = new SpeechRecognitionEngine(); 

 /// Recognize from default Microphone

 sre.SetInputToDefaultAudioDevice();            

 /// Grammar with the words you want to recognize

 GrammarBuilder grammarBuilder = new GrammarBuilder();

 grammarBuilder.Append(

     new Choices("start", "stop", "pause", "restart", "go back"));

 Grammar customGrammar = new Grammar(grammarBuilder);

 sre.UnloadAllGrammars();

 sre.LoadGrammar(customGrammar); 

/// Handler to be run in each recognition

  sre.SpeechRecognized +=     

      new EventHandler<SpeechRecognizedEventArgs>

         (sre_SpeechRecognized); 

 /// Do not stop recognizing

 sre.RecognizeAsync(RecognizeMode.Multiple);

You have to program a handler which will be used to handle the recognized text in the way that most pleases you.

void sre_SpeechRecognized(object sender,

    SpeechRecognizedEventArgs e){

      /// Store recognized text

      txbDebug.Text += e.Result.Text + "\r\n";

}

You can download my first Speech application to see it working