none
Additional languajes for KFW (kinect for windows) speech recognition SDK v1, When? RRS feed

Answers

  • We just announced that we will be adding a number of languages and locales in our 1.5 release at the end of May.

    We are also hard at work on our 1.5 release, which will be available at
    the end of May.  ...
    Also included in our 1.5 release will be four new languages for speech
    recognition – French, Spanish, Italian, and Japanese.  In addition, we
    will be releasing new language packs which enable speech recognition for
    the way a language is spoken in different regions: English/Great
    Britain, English/Ireland, English/Australia, English/New Zealand,
    English/Canada, French/France, French/Canada, Italian/Italy,
    Japanese/Japan, Spanish/Spain and Spanish/Mexico.

    http://blogs.msdn.com/b/kinectforwindows/archive/2012/03/26/what-s-ahead-a-sneak-peek.aspx

    Tuesday, March 27, 2012 3:45 PM
  • The English Acoustic model that ships with Kinect is optimized for the Kinect microphone array and will provide the best results.

    For the other locales, you can use the default acoustic model shipped with the V11 Microsoft.Speech platform, available here:

    http://www.microsoft.com/download/en/details.aspx?id=27224

    The results will not be as good as the Kinect specific acoustic model, though; you should expect a dramatic performance difference in recognition accuracy.

    Also, please remind to set the Kinect mic array gain to 3, check out the readme file for details.


    - hope this helps - Mauro.

    Tuesday, February 7, 2012 7:14 AM

All replies

  • I am also interested in additional languages!
    Wednesday, February 1, 2012 8:26 PM
  • I'm interested in german and russian languages.
    Wednesday, February 1, 2012 8:27 PM
  • Daniel, why not use the windows SAPI api? It has german built in.

    Clever, witty forum signature.
    Wednesday, February 1, 2012 10:53 PM
  • SAPI has also english built in, so why does Kinect SDK includes its own speech recognition for English language?

    Thursday, February 2, 2012 5:53 AM
  • SAPI has also english built in, so why does Kinect SDK includes its own speech recognition for English language?


    Given that Kinect's method of doing voice recognition is identical to SAPI's, I would guess that it actually just uses SAPI for its recognition (I think, do not know this!).

     

    If you look at the SAPI C# and C++ documentation, you might uncover some things that are useful to you. Specifically, this:

    http://msdn.microsoft.com/en-us/library/system.speech.recognition.grammarbuilder.culture.aspx


    Clever, witty forum signature.
    Thursday, February 2, 2012 6:01 AM
  • As far as I know Kinect does not uses System.Speech... but instead Microsoft.Speech.... that API is similar but has difirence.
    And as I said it has own recognition pack for English language...

    Thursday, February 2, 2012 6:04 AM
  • In that case,

    http://msdn.microsoft.com/en-us/library/microsoft.speech.recognition.grammarbuilder.culture.aspx


    Clever, witty forum signature.
    Thursday, February 2, 2012 6:14 AM
  • According to feature list: "Provide the latest Microsoft Speech components and an updated English Language Pack for improved language recognition accuracy.". This english language pack is not the same that comes with Speech SDK.
    So the question is still not answered - do we get additional language packs?
    Thursday, February 2, 2012 8:30 AM
  • The English Acoustic model that ships with Kinect is optimized for the Kinect microphone array and will provide the best results.

    For the other locales, you can use the default acoustic model shipped with the V11 Microsoft.Speech platform, available here:

    http://www.microsoft.com/download/en/details.aspx?id=27224

    The results will not be as good as the Kinect specific acoustic model, though; you should expect a dramatic performance difference in recognition accuracy.

    Also, please remind to set the Kinect mic array gain to 3, check out the readme file for details.


    - hope this helps - Mauro.

    Tuesday, February 7, 2012 7:14 AM
  • "The English Acoustic model that ships with Kinect is optimized for the Kinect microphone array and will provide the best results."
    So when "Kinect Acoustic model" for other languages will be available?

    Tuesday, February 7, 2012 8:28 AM
  • Left past 15 days after post the original question an no answer. It's frustrating how a simple is to misunderstand the original thread.

    We just want to know if there are additional language packs for the Kinect Speech Recognition. And I mean as good as the original English that ships with the Kinect SDK v1.

    Why am I asking this?, because as the promotional title befoore the SDK v1 said: "We are changing the world" or sort of. Well... I find kind of fantastic the Kinect as an extension for interacting with PC. But, I cannot change the world Just In English. If I want to create apps for changing the world, I really need a languaje that my customers can understand, so if I'm gonna buy the Kinect for windows, I need all the tools so I can develop my programs to interact with people (by the way my petition is for the spanish language pack). So if I buy a Kinect for windows sensor, will be when all the features can be used. Not only the Skeletal Tracking, Speech is the other half for a complete NUI software.

    Want us the help you Change the world?, don't just give us a pick, we need a shovel too.

    So... When will be additional (and tested languaje packs like English) for the Kinect Speech SDK v1?

    Wednesday, February 15, 2012 10:15 PM
  • One more nail in Kinect Audio сoffin. Microsoft Speech Platform (Microsoft.Speech) does have Russian language recognizer (super!), but there is no dictation. SAPI (System.Speech) does have dictation, but it does not have russian recognizer.

    Any ideas how to do a russian recognition with free grammar?

    Saturday, February 18, 2012 12:49 PM
  • We just announced that we will be adding a number of languages and locales in our 1.5 release at the end of May.

    We are also hard at work on our 1.5 release, which will be available at
    the end of May.  ...
    Also included in our 1.5 release will be four new languages for speech
    recognition – French, Spanish, Italian, and Japanese.  In addition, we
    will be releasing new language packs which enable speech recognition for
    the way a language is spoken in different regions: English/Great
    Britain, English/Ireland, English/Australia, English/New Zealand,
    English/Canada, French/France, French/Canada, Italian/Italy,
    Japanese/Japan, Spanish/Spain and Spanish/Mexico.

    http://blogs.msdn.com/b/kinectforwindows/archive/2012/03/26/what-s-ahead-a-sneak-peek.aspx

    Tuesday, March 27, 2012 3:45 PM
  • Why no german speech recognition?
    Tuesday, March 27, 2012 4:59 PM
  • We want to make sure that the languages packs that we release are at a high quality bar, and put in a ton of test time to make sure that they'll meet developer and end user expectations.

    German has been a little harder than some of the other languages, and the initial German recognizers that we trained on the Kinect for Windows pipeline exhibited high "False Activation" rates which would translate to commands being raised to the application which were not intended by the user.

    Additional language packs are definitely in our roadmap, and Germany is an important market, but we do not have a timeline further out than the 1.5 announcement that we can share at this point.



    Wednesday, March 28, 2012 6:51 AM
  • I thinkit's worthwaiting until May. I believe Speech recognition is the second half of the meaning of NUI. 

    Thanks, I will spect it with patience.
    Saturday, March 31, 2012 5:32 PM