Menu

Training speech to text

Help
2018-09-26
2018-12-18
  • Jose Sanchez

    Jose Sanchez - 2018-09-26

    Good morning:
    I have a question, how many people and hours of audio are needed to train a good acoustic model?

     
  • Jose Sanchez

    Jose Sanchez - 2018-09-27

    Thanks, another question, can I adapt a 16k acoustic model with 8k audio?

     
    • Nickolay V. Shmyrev

      No, it does not make much sense.

       
  • Jose Sanchez

    Jose Sanchez - 2018-10-25

    Hello Nickolay, is there any standard text to start with audio recording for sphinx training?

     
    • Nickolay V. Shmyrev

      These days you should never record speech specifically for training, you'd better take existing recordings.

       
  • Jose Sanchez

    Jose Sanchez - 2018-10-25

    Thank you, and what hardware requirements are needed for the training?

     
    • Nickolay V. Shmyrev

      A good server for a beginner ASR developer is

      Intel® Core™ i7-6700 Quad-Core
      64 GB DDR4 RAM
      2 x 500 GB SATA 6 Gb/s SSD
      GeForce® GTX 1080

       
      • Jose Sanchez

        Jose Sanchez - 2018-10-30

        Thanks, do not know if an acoustic model is available in 8000 Hz for the Spanish language?

         
        • Nickolay V. Shmyrev

          Not yet

           
      • Jose Sanchez

        Jose Sanchez - 2018-11-06

        Hi Nickolay, I would like to know what is the calculation made by the software at the time of transcription and training, because if a mathematical calculation is made, maybe a Geforce GTX is not the ideal model, since this model is focused on graphics for video game. I remain attentive to your response, thank you in advance.

         
        • Nickolay V. Shmyrev

          made by the software at the time of transcription and training

          Matrix multiplication.

          Geforce GTX is not the ideal model, since this model is focused on graphics for video game

          No worries about that.

           
          • Jose Sanchez

            Jose Sanchez - 2018-12-18

            Thanks Nickolay, We just acquired a graphics card, how do I get CMU-Sphinx to occupy the gpu? Is there a configuration?

             
            • Nickolay V. Shmyrev

              cmusphinx does not support gpu, other modern toolkit do.

               
              • Jose Sanchez

                Jose Sanchez - 2018-12-18

                ok, thanks and what do you recommend to recognize voice using gpu?

                 

Log in to post a comment.

Want the latest updates on software, tech news, and AI?
Get latest updates about software, tech news, and AI from SourceForge directly in your inbox once a month.