论文部分内容阅读
为了在恶劣丢包伴随高误码率的无线信道上进行实时、高质量的语音通信,提出了一种基于信源信道联合编码设计的“桶形移位联合叠加多数判决”语音编码传输算法。该算法在编码端将语音编码输出的语音参数非等重保护,然后分组进行BCH编码并经过桶形移位联合叠加合路输出。在解码端从接收到的语音数据包中依次提取出每组数据进行Berlekamp译码并根据是否在译码能力置当前组状态并缓存译码结果,遍历各组译码数据进行叠加多数判决,最后结合语音特性对线谱对参数进行差错后处理。该算法在无任何算法延时且消耗较小带宽的条件下抗丢包性能、纠错性能和端到端合成语音质量均优于传统“RS编码深度交织”算法,剩余误码率平均降低84.36%,合成语音MOS(meanopinion score)分平均提高38.86%。
In order to carry out real-time, high-quality voice communication on the bad channel with high bit error rate, this paper proposes a new method based on the combination of source channel coding and “Barrel Shift Joint Superposition Decision” algorithm. The algorithm encodes the speech parameters of the speech coding unequal-weight protection at the encoding end, and then carries out BCH encoding by grouping and outputs by combining the barrel shift and the superimposed combination. At the decoding end, each group of data is sequentially extracted from the received voice data packet to perform Berlekamp decoding. According to whether the current group state is set in the decoding capability and the decoding result is buffered, each decoding data is traversed to superpose the majority decision. Finally, Combining the phonetic characteristics of line spectrum parameters error post-processing. The proposed algorithm has better anti-packet loss performance, error correction performance and end-to-end synthesized voice quality than the traditional “RS coded depth interleaving” algorithm without any algorithm delay and consumes a small bandwidth. The average residual error rate Decreased by 84.36% and mean speech score increased by 38.86% on average.