Model Not able to recognize wav files with Noise

Acoustic Model Discussions

User: champ
Date: 12/18/2010 12:56 am

Views: 4768
Rating: 8

I adapted the WSJ model with some of the sounds(noiseless) and now when i record some sounds through my microphone then the sounds dont get recognised(most of the time i get null pointer execption and else i get some random result)... While all the sounds that i used to make the acoustic model are being recognised properly....

Is there a special way to record sound files??

Is there a way to reduce noise in the wav files??

or is there some other problem??

Additional Details(M Working on Sphinx 4,I use Ubuntu Hardy Version,2 GB of Ram)

Thanks In Advance!!!! :D

--- (Edited on 12/18/2010 12:56 am [GMT-0600] by champ) ---

Re: Model Not able to recognize wav files with Noise

User: kmaclean
Date: 3/21/2011 12:11 am

Views: 2068
Rating: 6

>I adapted the WSJ model with some of the sounds(noiseless)

do you mean speech sounds or non-speech sounds?

>Is there a special way to record sound files??

assuming you are recording speech sounds, then you need to make sure that you are recording speech with the same sampling rate as what the WSJ model was trained with.

--- (Edited on 3/21/2011 1:13 am [GMT-0400] by kmaclean) ---

Previous • Next •


Username	Password