Using VTLN matrices for rapid and computationally-efficient speaker adaptation with robustness to first-pass transcription errors.

Shakti Prasad Rath, Srinivasan Umesh, Achintya Kumar Sarkar

Browse the full Interspeech paper archive.