A Speech Signal Compression Algorithm Based On Human Auditory System And Wavelet Transform

As the rapid development of communication and computer network, the speechsignal compression technology is widely applied in many fields, such as mobilecommunication, satellite communication, multimedia and IP phone. The goal ofthis technology is to decrease the transfer bit-rate and to keep the voicegood quality at the same time, so to get more channels for transferring data.In this dissertation, a new speech signal compression algorithm based onwavelet transform(WT) and human auditory system(HAS) is proposed. Thisalgorithm using filter-banks constructed by double-quadrature WT whichsatisfies the HAS of human ears. Quantization threshold for output signal ofall the sub-band filters is established based on the loudness apperceptionand concealment features. Compression rate could be selected freely byadjusting the threshold for necessary performance of recovered speech, whichcan be judged by subjective method (the HAS of human ears) and objectiveparameters such as signal-to-noise rate(SNR). Experiments on evaluation showthat the algorithm is very efficient, it can provide good recovered qualitywith rather higher compression rate.