8 kbits/s speech coder with time scale modification and adaptive vector quantization
Tian-Min Liu, Lei Ye, Jian-Xiong Bai, Zheng Hu · 1st IASTED International Symposium on Signal Processing and its Applications · 1987
In this paper, we present an 8 kbits/s speech coder with time domain harmonic scaling (TDHS) of speech and a sequentially backward adaptive vector quantizer (AVQ-SB). The speech signal sampled at 8 kHz, is first compressed with a 2:1 TDHS, and the compressed speech is then quantized by an adaptive vector quantizer at 2 bits/sample. The motivation for using TDHS is two-fold. First, it can exploit the redundancies of speech in pitch structure and local stationarity, which are not used by VQ. Secondly, with a TDHS compression process, the equivalent sampling rate is reduced by a factor of 2, which will lessen the computation load of the VQ coding process. There are also two reasons for which we use AVQ-SB. First, as TDHS will bring delay. If we use frame adaptive VQ, it may introduce more delay. Secondly, adaptive vector quantizer (AVQ) can take advantage of speech amplitude nonstationarity, which may also improve the performance. Computer simulations results which demonstrate the coding system's performance are included.