Giters

plison / opendial

A generic Java toolkit for building dialogue systems

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Add options to specify custom acoustic model and dictionary in Sphinx ASR plugin

bryandeng opened this issue · comments

commented

The plugin requires one parameter:

  • "grammar" is the path to the recognition grammar in JSGF format

I think it's also a good idea to provide parameters for acoustic model and dictionary (maybe also via domain specification), instead of hard-coding these variables in SphinxASR.java.
(For example, in our project we need to use multiple models and dictionaries and switch between them. )

commented

OK, I've now updated the Sphinx plugin in order to use the acoustic model and dictionary as parameters. I've also added the possibility of using a statistical language model instead of a grammar. So now the plugin takes three parameters instead of one: (1) the path to the acoustic model, (2) the path to the dictionary file, and (3) the path to the grammar file or the path to the language model file.

Let me know if that works for you!