Hi,

     I followed the steps given in voxforge for the dialing application of speech recognition. Created the triphone state model. Recorded wave files from the general headphone set and all wavefiles are not recorded with same pitch, same loudness nor followed the prosodic. Even then i am able to recognition with good accuracy. Thanks to voxforge it helped me a lot.

Thank You

Ravi Kumar

voxforgeravi-26092011.tgz