Speech Recognition starting and ending events?

Question

I have simple commands like a)Take me to home screen b) go to next window etc. for speech recognition.

I just want to know which events raise when speech is started and ended.

I am considering the "speech detected" event when user starts speaking.

I am not clear which event fires when speech(user stops speaking) is stopped (tried using recognized/rejected/recognize completed events but unable to analyse when user speech ends exactly)

Can anyone help in this regard.

Answer 1

can anyone please clarify me?

Answer 2

When the SpeechRecognitionEngine detects a input that can be identify as speech raises the SpeechDetected Event

When the engine finalizes an asynchronous reconition operation raises the RecognizeCompleted event.

I suggest you to see the RecognizeCompletedEventArgs Class ;)

Answer 3

Hi,

I observed the RecognizeCompleted event raises if Speech is Recognized/Rejected/Hypothesized with the below code but the speech listening occurs only once and not continuous as it is in Single Mode.

sre.RecognizeAsync(RecognizeMode.Single);

 sre.RecognizeCompleted += new EventHandler<RecognizeCompletedEventArgs>(sre_RecognizeCompleted);

Is there a any way that I could get the same functionality in RecognizedMode.Multiple mode. Because with this mode, whether the speech is recognized/rejected/hypothesized, the RecognizeCompleted event is not firing.

My Requirement is 

a) the speech listening should be continuous

b) I should know when the user speech ends

suppose if my first command is , "go to home page", whether it is recognized/rejected/hypothesized i should know that user stopped speaking before the user goes to the next command.

Please help me how to achieve this.

Regards,

Bharat.

Answer 4

Hi, Can anyone please help me out.

Answer 5

The functionality of the Speech API's are not supported in these forums. You may want to ask over in one of the .Net forums and isolate the question to be generic and not Kinect specific. The Kinect API's only provide the Speech API's the audio stream data.

From the description of the API:

http://msdn.microsoft.com/en-us/library/system.speech.recognition.recognizemode.aspx
- Specifies that recognition does not terminate after completion.

Based on that, you shouldn't get a RecognizedCompleted event since the SpeechRecognizer doesn't stop.

Have you tried the any other samples? The ShapeGame sample uses RecognizeMode.Multiple. Does it work? You should be getting SpeechRecognized or SpeechRecognitionRejected events. You could also try the same functionality with a regular microphone.

http://msdn.microsoft.com/en-us/library/system.speech.recognition.speechrecognitionengine.loadgrammar.aspx