Menu ▾ ▴

Data preparation for sphinx train

Help
Hendrik Sirges
2006-07-03
2012-09-22

  • Hendrik Sirges

    Hendrik Sirges - 2006-07-03

    Hi,

    i'm currently trying to train my own acoustic model with sphinx train.

    i have some questions about howto use sphinxTrain

    1) my phonelist contains phones like
    9~
    @
    E
    e:
    Are there problems using upper/lowercase characters and special characters?

    2) What should a trainscript file (as described in the SphinxTrain doc under "DataPreparation=>3"). does it contain filenames or the transcript itself.

    3) I'm confused in general if i should follow the SphinxTrain documentation or should use the perl scripts provided with sphinxTrain. is the tinydoc
    the right place to start?

    4) can i use raw audio when using the perl scripts or do i have to convert to wav-format?

    thanks in advance for your answers

    greetings,

    Hendrik

     

Log in to post a comment.