가산잡음환경에서 강인음성인식을 위한 은닉 마르코프 모델 기반 손실 특징 복원

초록

This paper describes a robust speech recognition technique by reconstructing spectral components mismatched with atraining environment. Although the cluster-based reconstruction method can compensate the unreliable components fromreliable components in the same spectral vector by assuming an independent, identically distributed Gaussian-mixture processof training spectral vectors, the presented method exploits the temporal dependency of speech to reconstruct the componentsby introducing a hidden-Markov-model prior which incorporates an internal state transition plausible for an observed spectralvector sequence. The experimental results indicate that the described method can provide temporally consistent reconstructionand further improve recognition performance on average compared to the conventional method.

키워드