単語グラフ統合を用いた種々の雑音環境下での音声認識

摘要

本論文では，種々の雑音環境下での音声認識の性能向上のため単語グラフ統合を提案する．具体的には，混合連続分布型HMM（CMHMM）と離散混合分布型HMM（DMHMM）の2種類の音響モデルを使用することで単語グラフ統合を行う．DMHMMを用いたシステムが突発性雑音下での認識性能を大きく向上させることが，これまでの研究で確認されている．さらに，CMHMMを用いたシステムが高SNRの環境雑音において良い結果を示すことも確認されている．このようにCMHMMとDMHMMは異なる誤り傾向を持つ．上記のような理由からCMHMMとDMHMMを用いたシステム統合法を利用する．認識実験の結果，提案法はROVERやベースラインの認識結果と比べて良好な結果を得た．特に複合雑音下において認識性能が大きく向上した．%In this paper, we propose a new word graph combination (WGC) approach for speech recognition under various noise conditions. Specifically, we developed a WGC technique in which both continuous-mixture hidden Markov models (CMHMMs) and discrete-mixture hidden Markov models (DMHMMs) are being used as acoustic models. It has been previously verified that a DMHMM-based system can ensure significant improvements in the speech recognition performance under impulsive noise conditions. We also showed that the CMHMM-based system indicated better performance in environmental noise conditions at high SNR. Thus, the CMHMM and the DMHMM exhibit different error trends. On the grounds of the above mentioned findings, we adopted a system combination approach in which both a DMHMM and a CMHMM are used. In the speech recognition experiments we performed, the proposed system showed better performance than the ROVER-based system or the baseline system. In particular, this new system showed comparatively higher performance under mixed noise conditions.