The effects of room acoustics on MFCC speech parameter.

摘要: Automatic speech recognition systems attain high performance for close-talking applications, but they deteriorate significantly in distant-talking environment. The reason is the mismatch between training and testing conditions. We have carried out a research work better understanding of effects room acoustics on feature by comparing simultaneous recordings close talking distant utterances. characteristics two degrading sources, background noise reverberation are discussed. Their impacts spectrum different. affects valley while causes distortion at peaks pitch frequency its multiples. In situation very few data, we attempt to choose efficient compensation approaches spectrum, subband or cepstrum domain. Vector Quantization based model used study influence variation vector distribution. results speaker identification experiments presented both data.