摘要
运用基于离散Teager能量算子的浊音端点检测技术,将浊音端点及其之后的若干帧作为丢帧补偿的增强保护帧.在研究窄带语音帧间独立编码算法iLBC的基础上,对G.722.2编码标准进行改造,使之能够对宽带输入语音进行类似的帧间独立编码,以增强其抗丢帧鲁棒性.针对不同的丢帧率,对不同数目的增强保护帧进行帧间独立编码,以编码速率的略微提高增强对丢帧的鲁棒性,因此本算法既有iLBC的鲁棒性,又有一般CELP编码的速率自适应性.实验结果表明,相对于G.722.2编码标准附件I补偿方法,本算法可以提高丢帧补偿性能,在浊音端点丢失率较高的情况下的效果更好.
The detection technique for voiced onset based on discrete Teager energy operator was presented,then,the voiced onset and some frames following it were encoded specially as the enhanced protection frames.Based on the research of iLBC with frame-independent coding for narrow-band speech,G.722.2 was modified to code frame-independently for wideband speech to improve its robustness against frame error.According to different frame error rates,enhanced protection frames of different numbers were encoded frame-independently,and the robustness was gained at the cost of a slight increase in bit-rate.The algorithm possesses the robustness of iLBC and also bit-rate adaptability of codecs based on common CELP model at the same time.Experiment results indicate that,compared to the frame error concealment in G.722.2 Appendix I,the proposed algorithm can improve the concealment performance,and enhances concealment effect especially when the loss rate of voiced onset frames is higher.
基金
国家自然科学基金(60971098)资助
