Menu

LDA / MLLT -reg

Help
2011-08-04
2012-09-22
  • Madhav Kishore

    Madhav Kishore - 2011-08-04

    this is from ref
    http://cmusphinx.sourceforge.net/wiki/ldamllt**

    1.Specifically, you have to train a basic model in order to train each feature
    transformation, then retrain the model with the transformation applied to
    the input features
    . This has to be done for each feature transformation

    2.This means that it's necessary to define the set of acoustic classes on
    which they are trained.

    I am not clear...kindly explain what they mean

    **

     
  • Nickolay V. Shmyrev

    Specifically, you have to train a basic model in order to train each feature
    transformation, then retrain the model with the transformation applied to the
    input features. This has to be done for each feature transformation

    First LDA transform is trained on stage 01, then MLLT transform is trained on
    stage 02

    This means that it's necessary to define the set of acoustic classes on
    which they are trained.

    Current implementation separates CI phones. You define them during database
    setup.

     

Log in to post a comment.