基于MELP的水下实时语音通信机的研究与实现

Research and implementation of underwater acoustic real-time voice communication prototype based on MELP

根据水声通信的复杂信道特性,对混合激励线性预测语音编码MELP算法和低密度奇偶校验码LDPC的结合进行了研究,设计了基于ARM+DSP的多载波水声实时语音通信样机。经水池实验测试,系统表现出良好的实时性和语音可懂度性能,可分辨出不同说话人的声音。

Taking the complexity of the underwater acoustic communication channel characteristics into account, the combination of mixed excited linear prediction（MELP） and low density parity check（LDPC） were studied and a multi-carrier underwater acoustic real-time voice communication prototype based on ARM＋DSP was designed. By the expreimental test in pool, the proposed system showed a good real-time and voice intelligibility performance, and could distinguish the voice of different speakers.