水下双工语音通信系统的设计与实现

蛙人语音通信是水声领域研究的热点问题。利用多帧联合处理以及矢量量化技术实现了1.2kbps 速率的混合激励线性预测语音压缩编码，为适应水声信道的特点，采用正交频分复用技术对信源进行调制，并加入了同步、信道估计、RS纠错码等相关技术，以TMS320DM642和TLV320AIC23B等芯片为核心搭建硬件平台，设计并实现了一套水下实时双工语音通信系统。水池试验表明，在20 m距离上，合成语音清晰连贯并可分辨出说话人为谁，系统性能稳定。

Design and implementation of the underwater duplex speech communication system

Underwater frogman speech communication is always a hot issue in underwater acoustic field. In this paper a 1.2kbps bit rate Mixed Excitation Linear Prediction (MELP) speech compression algorithm is implemented through multi-frame joint and Vector Quantization (VQ) technology. In order to adapt to underwater acoustic channel, the Orthogonal Frequency Division Modulation (OFDM) is used to modulate the source bits, together with synchronization, channel estimation and Reed-Solomon (RS) error correcting code techniques. The hardware platform of this speech communication system relies on the chips DM642 and AIC23 and some auxiliary circuits for SDRAM and FLASH. The system actualizes duplex speech signal acquisition and transmission and tank-testing results prove that the synthesized speech is clear and understandable at the distance of about 20 meters and the system has a stable performance.