Channel and noise compensation for text dependent speaker verification over telephone

The performance of text dependent, short utterance speaker verification systems degrades significantly with channel and background artifacts. The authors investigate maximum likelihood and adaptive techniques to compensate for a stationary channel and noise. Maximum likelihood channel and noise compensation was introduced by Cox and Bridle (1989), and has been shown to be effective in many other speech applications. For adaptive estimation, a Bussgang like algorithm is developed which is more suitable for real-time implementation. These techniques are evaluated on a speaker verification system that uses the nearest neighbor metric. The results show that for telephone speech with channel differences, channel compensation can provide substantial performance improvement. For un-cooperative speakers, background compensation resulted in a 35% improvement.