Sounds great, thanks for the submission.

Yes, please do contribute as much audio as you can!

The target amount of audio for release 1.0 of the VoxForge Speech Audio Repository is 140 hours of speech (why? this is the same amount of audio used by the Sphinx Acoustic Models used by the Sphinx Group Speech Recognition Engines.)

thanks,

Ken 

--- (Edited on 10/12/2006 12:43 pm [GMT-0400] by kmaclean) ---

Thanks :)

The instructions were correct, it DID take quite a bit of time.  I will try to add more to the project.

The only thing that caused me problems was that I was unable to determine my "dialect" according to the instructions... there really isn't a category that fits, so I chose "General American English".  I live in Norfolk, VA, was born and raised in Richmond, VA.  Most would not categorize my speech as "southern", but it isn't "Mid-Atlantic" (even though Virginia *is* Mid-Atlantic it is not included in the definition offered on the site).

 I suppose it would be better for your reviewers to decide which categories in which to place the submitted speech.

--- (Edited on 10/12/2006 5:18 pm [GMT-0500] by crxssi) ---

 

Sorry for the late reply, I was relying on receiving email notices for new posts, but I must have missed yours.

My recommendation is to try to classify it yourself as best as you can. 

As we get more speech data, we will be in a better position to classify it properly, since we will have more audio to compare it against.

Note: in addition to ensuring we have coverage of all english dialects when we create Acoustic Model, another purpose of categorizing speech by dialect is so we can create specialized Acoustic Models targeted to certain dialects or regions.  We would do this to reduce the size of the Acoustic Model, and to improve recognition accuracy - because there would be less variation in the sounds that the Acoustic Model was trained to recognize.  I tried to limit the number of dialects because of this - but if you need to add a new dialect, by all means do so.

Ken

 

--- (Edited on 10/18/2006 11:41 am [GMT-0400] by kmaclean) ---