Nonspecific Speech Recognition Based on HMM/LVQ Hybrid Network
Shuling Liang, Wang Chaoli, Jiaming Du · 2009
A novel method of speech recognition, which is based on HMM/LVQ1-LVQ2, is proposed in this paper. First, the MFCC, DeltaMFCC and DeltaDeltaMFCC extraction algorithms are introduced, then these coefficients are normalized by HMM-based Viterbi method, after that, the normalized feature sequences are got. The recognition is first to learn coarsely by using LVQ1 and then to learn finely by LVQ2. Finally the result is given, which shows the proposed algorithm improves the recognition rates effectively, in comparison with HMM used alone or LVQ1-LVQ2 hybrid network recognition, especially for nonspecific speech.