Packet Loss Concealment Of Speech Transmissions Based On Compressive Sensing

Recent years, with the rapid development of Internet, VoIP (Voice over Internet Protocol) has attracted more attention and it is expected to exceed the current public switched telephone networks in the near future. Real-time service in the IP network is always facing the following problems:jitter, delay and channel errors. These problems lead packet loss in the receiver, and make severe speech quality degradations.Therefore, based on the concept of CS (compressive sensing), this paper proposes a receiver-based PLC scheme-CS-PLC. This scheme can be widely used in speech transmissions system, processing the anti-packet loss with low algorithm complexity and effective repair ability. Furthermore, this technique does not need the sender to participate, and the reconstructed speech will be more natural. This scheme can be used in many speech communication systems to improve speech transmission quality with strong applicability.The CS principle and the CS-PLC algorithm are introduced first, then the key technologies of CS-PLC are discussed, and the performance of the algorithm is evaluated in the end. How can CS-PLC algorithm be applied to speech transmission for improving speech transmission quality is the focus of this study. Test results have shown that, the CS-PLC algorithm is an effective packet loss concealment method. When it is used with the conventional speech codec, speech transmission quality can be improved significantly. Moreover, as a developing technology, the CS-PLC algorithm also needs to be further improved.