Hello,

I am looking to detect American-English spontaneous conversation from a VoIP output.

Are there any pre-built hmm training models already available which can be directly used for recognition through HTK!

If not, then where can I find free text-annotated speech corpus which will allow for training and recognition in my above mentioned case.

Thanks!

--- (Edited on 1/18/2011 2:33 pm [GMT-0600] by ) ---