I have previously trained acoustic models using SphinxTrain. I want to use
these models as the initial means, variances, etc. in a new training with new
data. I believe this is called bootstrapping in HTK. Is there a way to do this
in SPHINX?
Thanks.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
You can use model reestimation tools with existing model. There is actually no
difference how to use bw and norm. You can get details on tool arguments from
the documentation
I have previously trained acoustic models using SphinxTrain. I want to use
these models as the initial means, variances, etc. in a new training with new
data. I believe this is called bootstrapping in HTK. Is there a way to do this
in SPHINX?
Thanks.
You can use model reestimation tools with existing model. There is actually no
difference how to use bw and norm. You can get details on tool arguments from
the documentation
http://www.speech.cs.cmu.edu/sphinxman/scriptman1.html
You can find the examples of tools invocation in the scripts in
Sphinxtrain/scripts_pl.