German

Nested
Adapting existing or training a new model ?
User: CMikado
Date: 3/26/2012 11:50 am
Views: 6495
Rating: 8

Dear Voxforge-Team,

I am trying to use your well appreciated work on the German Voxforge Acoustic Model for Sphinx to integrate it into the PocketSphinxDemo for Android. Unfortunately I was told that it won't work with a telephone sample rate and I have to train a 8khz model with German audio data. So my question is : Is there a way to adapt the existing Sphinx-Model in a way that it also accepts 8khz audio ? Using that tutorial : http://cmusphinx.sourceforge.net/wiki/tutorialadapt

 

Or do I have to create and train a completely new one with the help of the provided VoxForge audio data ?

Thanks for the help.

Kind regards !

Re: Adapting existing or training a new model ?
User: kmaclean
Date: 3/26/2012 9:35 pm
Views: 1375
Rating: 7

>Is there a way to adapt the existing Sphinx-Model in a way

>that it also accepts 8khz audio ? 

I don't know of an easy way to adapt a 16kHz-16bit acoustic model to a 8kHz-16bit acoustic model - best to train your own...

Re: Adapting existing or training a new model ?
User: fayre333
Date: 12/6/2017 4:08 am
Views: 1659
Rating: 0

Hey, did you have any success by training your own 8khz acoustic model? I'm working with a bluetooth headset so I guess I need an 8khz coustic model, too, but I'm totally new to all this and I don't know where to start to adapt or to train my own model.

PreviousNext