论文部分内容阅读
中远距离(>10 km)水声语音通信时,由于可利用带宽窄、复杂多变等不利因素对信息传输率的制约,语音编码速率应降到尽可能的低。利用水声信道传播时延大的特点,结合人耳听觉感知的特性,在深入研究混合激励线性预测编码(MELP)标准之后,提出一种语音编码速率可调节的变比特率语音编码算法。其平均码速率约600 bps,主观语音质量评估平均得分(PESQ MOS)约2.8分。对该编码算法性能进行了计算机仿真和海上实验验证。实验及仿真表明,在误码率不高于10~(-3)时,本算法表现良好且稳定,合成语音清晰可懂,易于辨认说话人。
For long-distance (> 10 km) underwater acoustic communication, the rate of speech transmission should be reduced as low as possible because of the constraints on the information transmission rate due to the narrow bandwidth and complex changes. According to the characteristics of human ear auditory perception, this paper proposes a variable bit rate speech coding algorithm with adjustable speech coding rate after studying the MELP standard. The average code rate of about 600 bps, subjective voice quality evaluation average score (PESQ MOS) of about 2.8 points. The performance of the coding algorithm was verified by computer simulation and sea experiment. Experiments and simulations show that the proposed algorithm performs well and steadily when the bit error rate is no higher than 10 ~ (-3), and the synthesized speech is clear and understandable, making it easy to identify the speaker.