Using Speech Recognition without Internet in Silver light applications

This article will focus on how to use  SpeechRecognizer in applications in order to enable Voice Recognition in offline mode. We’ll be building an application which will have a pre-defined user-customized grammar list and will be able to recognize the text from that grammar.

 

  • To start with exercise, create a Silverlight project of your choice, for this blog, we’ll be working with Blank App application.

 

 

  • Now, First thing we should do is open “WMAppManifest.xaml” and enable “ID_CAP_MICROPHONE” and “ID_CAP_SPEECH_RECOGNITION” under the “Capabilities” tag.

  • Next we’ll create a simple interface with one Application bar Button which will start the listening process and one TextBox. We’ll name the Textbox as “textbox”. Then create a click event_handler for the Button. Overall interface will look something like this
Designer View:

 

Xaml:
<Grid     <TextBlock Text="Speech Recognition without internet" Margin="9,-7,0,0" 
TextWrapping="Wrap" FontSize="50"/> 
    <TextBox x:Name="textbox" HorizontalAlignment="Left" 
Height="154" Margin="33,187,0,0" TextWrapping="Wrap" 
Text="Output box" VerticalAlignment="Top" Width="395"/> 
  
</Grid> 
  
  
<phone:PhoneApplicationPage.ApplicationBar    <shell:ApplicationBar Mode="Default" Opacity="1.0" 
IsMenuEnabled="True" IsVisible="True"        <shell:ApplicationBarIconButton Click="ApplicationBarIconButton_Click"
 IconUri="/Assets/AppBar/Microphone.png" Text="Listen"/> 
     
    </shell:ApplicationBar> 
</phone:PhoneApplicationPage.ApplicationBar>

Code:

  • C#
    SpeechRecognizer speechrecognizer; 
          
         public MainPage() 
         { 
             InitializeComponent(); 
             speechrecognizer = new SpeechRecognizer(); 
             string[] commands = { "Hello""This""is""easy"}; 
      
             speechrecognizer.Grammars.
    AddGrammarFromList("Commandset1", commands); 
         
         }
  • Now we come to the last step of the program that is writing the recognition code. We’ll be implementing all this in the event handler of button. Wait time will be specified, for the duration of which application will listen to speech before processing it. This time is specified by setting the value ofspeechrecognizer.Settings.InitialSilenceTimeout. In order to get the processed result, we need to create an object of class SpeechRecognitionResult  which we’ll be naming result. It will be initialized by speechrecognizer’s result. This is an Async process, so we’ll not only add await command before this but also make this whole function async. Refer to the snippet given below for implementation. Now all that’s left is seeing whether the Audio captured is from the defined grammar or not. For that, we’ll use SpeechRecognitionResult.TextConfidence property. This property can return High, Medium and Low, depending on the similarity with the text from grammar. We can implement any one of them, depending on how precise we want it to detect.

Code:

  •  C#
    private async void ApplicationBarIconButton_Click(object sender, EventArgs e) 
          { 
              speechrecognizer.Settings.InitialSilenceTimeout = 
    TimeSpan.FromMinutes(1.00); 
              SpeechRecognitionResult result = 
    await speechrecognizer.RecognizeAsync(); 
      
              if (result.TextConfidence == SpeechRecognitionConfidence.High
     || result.TextConfidence == SpeechRecognitionConfidence.Medium
     || result.TextConfidence == SpeechRecognitionConfidence.Low) 
              { 
                  try 
                  { 
                      textbox.Text = result.Text; 
      
                  } 
                  catch (Exception ex) 
                  { 
      
                  } 
              } 
          }
  • And that’s it, our basic application is good to go.
You can download the latest code from here.
Advertisements