I'm trying to convert a CI acoustic model trained in GMTK to sphinx3 format.
1) Will Sphinx3 run with a CI model, or do you need to go through the entire training sequence and create a CD model?
2) I've converted my model parameter files to the means, vars, mixw, and trans_prob ascii files described here: http://www.speech.cs.cmu.edu/sphinxman/scriptman1.html#4b
However, Sphinx3 expects binary files for means, vars, etc.
Is there a utility that will convert the ascii files described in the documentation (link above) to binary, or at least a description of the Sphinx3 binary acoustic model file format so that I can do the conversion myself?
Thanks
Alex
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi,
I'm trying to convert a CI acoustic model trained in GMTK to sphinx3 format.
1) Will Sphinx3 run with a CI model, or do you need to go through the entire training sequence and create a CD model?
2) I've converted my model parameter files to the means, vars, mixw, and trans_prob ascii files described here:
http://www.speech.cs.cmu.edu/sphinxman/scriptman1.html#4b
However, Sphinx3 expects binary files for means, vars, etc.
Is there a utility that will convert the ascii files described in the documentation (link above) to binary, or at least a description of the Sphinx3 binary acoustic model file format so that I can do the conversion myself?
Thanks
Alex
Format topics were discussed some time ago during work on conversion from HTK to sphinx3, see
https://sourceforge.net/forum/message.php?msg_id=4348648
https://sourceforge.net/forum/message.php?msg_id=4427328