Re: Sound Samples

On 11/27/03 6:39 AM, "MadJock" <[EMAIL PROTECTED]> wrote:
> 
> You say that the classification worked well with the exception of glide
> components.  May I ask what you were trying to classify?  Was it consonant
> and vowel sounds for speech recognition?
> 
> In my application, it will not be necessary to decode anything - mine is
> only a classification exercise too.  Thanks for the info, and your
> suggestions.  I'll give them a try and see if it produces the results for
> me.  The mean frequency seems like a good starting point.
> 
> Any further ideas from anyone else would also be appreciated.
> 
> Thanks again
> 
> MadJock
> 

If what you're doing is speech-related, you should check out some of Yoshua
Bengio's older stuff (using both HMMs and ANNs, as well as a hybrid of the
two)...he's at the University of Montreal...

http://www.iro.umontreal.ca/~bengioy/


Hope this helps...



Fred.