Menu

Question about recording to create the language model for alphabets only

Help
pannam
2017-01-15
2017-01-17
  • pannam

    pannam - 2017-01-15

    Hi,
    I just want to create a language model for alphabets only. As previously I have recorded a sentence of 10- 15 seconds and created transcription etc accordingly. But now I want to detect only a single alphabets at a time how should I record it ? Do I record individual alphabets and create their transcription or the complete alphabets and its transcription?

     
  • Arseniy Gorin

    Arseniy Gorin - 2017-01-17

    You can record strings with alphabets like a b c ... . Acoustic model training should figure out the boundaries if you have enough sentences. If not, maybe better training individual characters and having transcripts for them. Both approaches are possible in fact

    However, if you want to decode 1 character in a run, then you just modify the language model accordingly (using grammar or a simple LM).

     

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.