基于AMR的网络语音处理算法研究与性能的仿真

发布时间：2018-01-26 07:24

本文关键词： IP网络电话丢包自适应多速率语音编码评价模型仿真　出处：《东北大学》2014年硕士论文　论文类型：学位论文

【摘要】：网络语音是因特网语音发展的代表标志,它也可称为因特网语音(VoIP-Voice over Internet Protocol)技术。VoIP是在IP分组交换网络的基础上,来实现语音通信的传输技术,由于价格便宜和功能种类多,所以得到了广泛地应用。尽管VoIP在成本节省和服务改进方面提供了各种优势,但部分由于质量问题,其推广一直不太顺利。事实上,在传统电话系统和新型的VoIP系统之间存在着若干基本差异,网络的流畅与堵塞必然引起语音信号的延时、抖动、丢包、回声等现象,影响VoIP网络通话质量的主要原因就是丢包现象。在接收端,经过语音解码后的语音实时回放的质量的损失和劣化就是由于语音包的丢失所带来的。论文通过自适应多速率(AMR)语音编码技术在丢包处理上的算法,在VoIP系统中的信源信道,通过自适应的联合编码来实现丢包问题的解决方法。这种算法就是采用动态的网络状况,在信道编码上选取在Reed-Solomon码的前向纠错体制,运用E-Model评价模型对语音质量进行评估,然后自主选择一种能够达到语音质量最好的信源和信道编码速率。本文设计了一种自适应算法,适用于VoIP系统的信源信道速率,采用了AMR语音编码技术和基于RS码的前向纠错信道编码。这种算法通过E-Model评价模型从丢包和延时两点来判断语音的质量,依据当前变化的网络状况自适应地选择出语音质量最优的信源与信道编码方案。对于这种算法,以VisualC++6.0平台进行算法的分析,分别在不同的网络状态下,求解出使得语音质量最优时的信源信道速率以及前向纠错编码方式,如果与假定的结果相符,那么就证明了它的有效性。实验在NS2仿真环境下进行,模拟网络环境的变化,在VoIP系统中,结合信源、信道速率自适应算法,并与VoIP系统中固定速率G.729的编码器、VoIP系统中信源速率自适应AMR的编码器相比较。试验表明,论文设计的这种算法对于丢包和延时两个问题的影响都不同程度的下降,尤其是在网络拥塞的时候它的效果更加明显。
[Abstract]:Network voice is the representative symbol of Internet voice development. It can also be called VoIP-Voice over Internet Protocol. VoIP is based on IP packet switching network. The transmission technology to realize voice communication is widely used because of its low price and wide variety of functions, although VoIP provides various advantages in terms of cost saving and service improvement. In fact, there are some basic differences between the traditional telephone system and the new VoIP system. The fluency and blockage of the network will inevitably cause the delay, jitter, packet loss, echo and other phenomena of the voice signal. The main reason that affects the quality of the VoIP network is the loss of the packet. The loss and deterioration of the quality of real-time playback after speech decoding is due to the loss of speech packets. This paper uses adaptive multi-rate AMR-based speech coding technology in packet loss processing algorithm. In the source and channel of VoIP system, the solution of packet loss problem is realized by adaptive joint coding. This algorithm adopts dynamic network condition. In the channel coding, the forward error correction scheme of Reed-Solomon code is selected, and the evaluation model of E-Model is used to evaluate the speech quality. Then we choose a source and channel coding rate which can achieve the best speech quality. In this paper, we design an adaptive algorithm, which is suitable for VoIP system. AMR speech coding technology and forward error correction channel coding based on RS code are adopted. This algorithm judges the quality of speech by E-Model evaluation model from packet loss and delay two points. According to the current changing network conditions, the source and channel coding scheme with the best speech quality is adaptively selected. For this algorithm, the algorithm is analyzed on the platform of VisualC 6.0. In different network states, the source channel rate and forward error correction coding method are obtained when the speech quality is optimal, if the results are consistent with the assumed results. The experiment is carried out in the NS2 simulation environment to simulate the change of the network environment. In the VoIP system, combining the source and channel rate adaptive algorithm. It is compared with the source rate adaptive AMR encoder in the fixed rate G. 729 encoder of VoIP system. The effects of this algorithm on packet loss and delay are reduced in different degrees, especially when the network is congested.
【学位授予单位】：东北大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN912.3

【参考文献】