Menu

acoustic model adaptation

Help
Pankaj
2016-04-25
2017-03-16
  • Pankaj

    Pankaj - 2016-04-25

    Suppose I have two speakers speaker 1 and speaker 2. I adapt an existing model for speaker1. If I readapt the model adapted for speaker1 using speaker2's voice, will the final model show improvement in the accuracy of both the speakers or just speaker 2 or speaker 1.

     
    • Nickolay V. Shmyrev

      Just speaker 2

       
  • Elpimous

    Elpimous - 2016-04-30

    Well , Nickolay, Are you sure ??
    I don't agree with you, but perhaps i'm wrong
    if I adapt model with my voice for 1h,
    adapt model,
    and adapt my adapted model with my girlfriend voice,
    i should have a small better accuracy for both of us, no ?

     
  • Pankaj

    Pankaj - 2017-03-09

    Yes Nickol is correct. However my observation is that while adaptation if we use data from multiple speakers together then the accuracy increases for all the speakers. If the adaptation is done one by one i.e the adaptation is first run for speaker 1 and then the process is repeated for speaker 2 then the accuracy will imrove only for speaker 2.

     

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.