Voxforge Spanish Model Released

As of today, CMUSphinx project is proud to present you the Spanish acoustic model. It was trained on the acoustic data to support both wideband recognition of microphone speech recordings and narrowband recognition of the telephone speech. Now you can build Spanish IVR using CMUSphinx tools.

You can find the model HERE.

Please try and report the accuracy results.

Also, note that there are models for many languages like German, Dutch, French or Russian. In the future we are going to expand the set of languages for your convenience.

10 Responses to “Voxforge Spanish Model Released”

  1. Prabhu says:

    Superb… I really appreciate this job.. and your valuable initiatives….

  2. Mosur Ravishankar says:

    Thanks for this. Will the location of the models for other languages mentioned (French, German etc.) be made available? Also, more details about the training data (e.g. how many hours of speech used) would be of interest.
    Thanks again.

  3. admin says:

    Hello Ravi!

    Other models are located in the same downloads folder on sourceforge.


  4. admin says:

    As for statistics, the metrics could be found on Voxforge. For spanish model it’s around 10 hours I think


  5. assanta says:

    I have never used sphinx before, normally I use comercial ASR.
    But now after test the google ASR, i can’t guess how they do. It’s no possible to have a so big grammar. And this is what I’m looking for.
    Is possible to do something similar with “sphinx” and “Voxforge Spanish Model Released”, or it has nothing to do?


  6. admin says:

    Dear Assanta

    It’s perfectly possible to follow Google path. You just need a passion to go through it.

  7. assanta says:

    Sorry, I didn’t explain very well
    I wanted to ask if it’s possible to do a continuous speech recognition without grammar, with “Sphinx” and “Voxforge Model”

  8. Fernando says:

    and how use that models with a sphinx 4 to spanish model an how generate de spanish model.. .S thk a lot.. :D