Kinect SDK - Voice Identification RRS feed

  • Question

  • I was wondering if there's a way (through the kinect SDK or otherwise) to be able to identify the voice of a person speaking to kinect.  I already know how to use voice recognition commands via the SDK, but I was wondering if I can recognize the voice of the person so in code I'd know who's controlling the kinect.
    Saturday, August 6, 2011 4:25 AM


All replies

  • The Kinect SDK Beta does not provide this functionality, but others might know better about any libraries available for voice recognition.

    Good luck.

    I'm here to help
    Saturday, August 6, 2011 7:19 AM
  • You could look into a library from dragon naturally speakings website or another open source project that involves dotnet for the extra functionality and mix it with kinect code. Eddys right if you use voice recogition  dotnet library from another website it might have the functionality your looking for. 
    Wednesday, August 17, 2011 6:09 PM
  • I believe the Xbox accomplishes this by using facial recognition. Once a person's face is recognized visually, you can get a vector to their position. Then, take the current audio stream, get a vector on the direction, and see which person it matches up with. I do not know of a way to do recognition on a voice sample along in the API. 
    Thursday, September 1, 2011 7:05 PM
  • Try installing the the windows speech sdk that has might have the functionality your looking for and combine that with the kinect drivers that recognizes it as a microphone and your in business to building your app. Kinect sdk byitself doesn't have this precise functionality. But when combined with the drivers from the kinect sdk you can build a custom app with dragon naturally speakings sdk or microsofts or with a library for open source dot net apps. I would listen to eddy or chris they both are right but i think eddy's is still the best option.
    Saturday, September 3, 2011 1:31 AM