一种新的中速率语音编码方案

Research on a New High-Grade Plan of Medial-Rate Coding for Speech

本文提出了一种新的混合型中速率语音编码方案,它充分考虑了语音(尤其是汉语)的特点、能量分布和人耳的听觉特性,把语音先分割成基带(300Hz～1kHz)和高频带(1～3.4kHz),然后分别采用波形编码(ADPCM)和参数编码(MELP)对基带和高频信号进行处理.实验结果表明,其信噪比较高,尤其是主观质量很好,可与24kb/s甚至32kb/s的ADPCM方案相比拟,并且它的高低频段信号可以同时进行运算,实时处理很客易实现。

In this paper, we propose a new mixed medial-rate speech coding method in which the voice haracteristic, the energy distribution and the ability of hearing are fully considered. The character of this scheme is that the peech is firstly divided into baseband signal (0.3～lkHz) and highband signal (l～3.4kHz), then wave coding (ADPCM) and arameter coding (MELP) algorithms are used. Simulation has proved that recovered speech has good naturalness, learness and identifiability. Its subjective quality is very good and may be comparable with 24kb/s even 32kb/s DPCM. In addition, baseband and highband signals can be operated in parallel, it is easy to realize this new scheme in real-time.