Menu

My first project - Speech recognition

Francesco
2019-03-20
2019-04-13
  • Francesco

    Francesco - 2019-03-20

    Hello,
    i have a list of wav file and i would like to transcribe in a text file (one to one), how can i do it?
    I would like to know how can improve the trascription and which are the main command to transcribing the text.
    I read in the website some guide, but i'm very confused.
    I need to use the following model https://sourceforge.net/projects/cmusphinx/files/Acoustic%20and%20Language%20Models/Italian/
    Someone can help me please?

    Regards
    Frank

     
    • Nickolay V. Shmyrev

      Sure, what have you tried already please?

       
      • Francesco

        Francesco - 2019-03-20

        Hi Nickolay,
        i download the following software: sphinxtrain, pocketsphinx and sphinxbase.
        I read that i need to normalize the wav file in 16000hz mono and now
        i'm in stuck. How can i use the italian model and inprove it? How can I start the first transcription from a wave file? Can you give me also some advice?It's my first experience with sphinx :/

        Cheers
        Frank

         
        • Nickolay V. Shmyrev

          Your tutorial is pretty staightforward to follow from start to end

          http://cmusphinx.github.io/wiki/tutorial

          if you stuck at particular problem, let us know!

           
          • Francesco

            Francesco - 2019-03-21

            Thanks Nickolay, i will try and i will let you know.

             
          • Francesco

            Francesco - 2019-03-22

            Hi Nickolay,
            I'm trying to read the tutorial, but I have some problems.
            When i download the file cmusphinx-it-5.2.tar.gz and i try to execute setup.sh nothing happen, why? My OS is Ubuntu.

             
            • Nickolay V. Shmyrev

              You have to edit that script according to your needs in order to run it.

               
  • Francesco

    Francesco - 2019-03-23

    Hi Nickolay,
    I tried to use the Italian model, but when i try to decode the file the transcription was wrong, can you give me some advices?i've read also the tutorial, but i have not experience with speech analytics :(

     
    • Nickolay V. Shmyrev

      Italian model is pretty basic, I will have to train it from scratch with more data. It is a long story, not sure it will go public soon.

       
  • Francesco

    Francesco - 2019-04-09

    Hi Nickolay,
    can you give me some advice to make a good train?

     
    • Nickolay V. Shmyrev

      I will be actually training Italian this months, I can share with you once done.

       
      • Francesco

        Francesco - 2019-04-13

        Great!Let me know when it will be ready :)
        Thanks

         

Log in to post a comment.