实时语音的自适应回放算法

由于丢包和延时抖动的引入而使网络传输的实时语音质量让人难以接受，目前对丢包和延时抖动提出了很多的解决方案．但是却很少把这两者结合在一起进行研究。本文提出了一种新的自适应回放算法，通过监测接收和回放队列，结合丢包的自适应恢复技术，达到语音高质量的连续回放。实验证明，该算法能在严格的平均回放延时条件下努力减小由于超时而引起的丢包，获得较好的重建语音质量。

Adaptive Playout Algorithm for Real-Time Packet Voice

The quality of real-time packet voice is unacceptable due to packet loss and delay jitter. Recently much effort is involved in packet loss and delay jitter. However, the tradeoff between delay jitter and packet loss is exploited few or no. In this paper, one new adaptive playout algorithm is proposed, which is able to play packet voice consecutively. By monitoring the receiver and playout queues, the algorithm adapts to delay jitter by increasing or decreasing playout delay in response to the thresholds of the length of queues. With adaptive recovery techniques for packet loss, the algorithm can obtain a high quality with few computational complexities. Experimental results validate that the adaptive playout algorithm can strive for minimizing packet droppings for late-arrival packets under a tightly bounded average playout delay with the acceptable speech quality.