论文部分内容阅读
基于计算机上的模拟,我们在波形编码的范畴内,用APC技术实现了一种速率低到4800bps的实时话音处理系统。该系统利用二重预测器消除话音信号中的多余度,井靠传送部分剩余误差极性信息和波形准周期搬移方法达到数码率的大压缩。用600余块大、中、小规模集成电路片子实现的系统硬件,实时工作话音不仅具有相当好的自然度,而且具有良好的清晰度。实验表明,它即使工作在高达70db(SL)的低频窄带环境噪声中和误码高达1.4×10~(-2)的信道条件,重建话音仍然清晰易懂。这就使得它特别适合于环境和信道条件都较差的场合进行通信。
Based on computer simulations, we implemented a real-time voice processing system at a rate as low as 4800bps using APC technology in the context of waveform coding. The system uses the double predictor to eliminate the redundancy in the voice signal and achieve the large compression of the digital rate by relying on the information of the residual error of the transmitted part and the quasi-periodic waveform migration. With more than 600 large, medium and small scale integrated circuit chips to achieve the system hardware, real-time voice work not only has a very good degree of naturalness, but also has good clarity. Experiments show that even if it operates in low-frequency, narrowband ambient noise up to 70db (SL) and up to a channel error of up to 1.4 × 10 -2, reconstructed speech remains legible. This makes it particularly suitable for communication where environments and channel conditions are poor.