随着计算机网络的快速发展，VoIP电话业务也得到广泛应用。这种基于IP网络的语音通话技术提供了方便廉价的交流途径，给通信行业带来了翻天覆地的变化。但随之而来出现的话音质量问题的解决，也迫在眉睫。WebRTC语音引擎是目前最先进的语音引擎之一，其中包含了实现语音信号处理的NetEQ模块，此模块很好地解决了VoIP电话业务出现的延迟、抖动和丢包问题。但由于NetEQ是专利技术，没有开放相应的开发文档，因此需要对NetEQ进行详细分析研究，以便能够更好地应用于VoIP业务中。本文研究的核心内容是NetEQ集成的自适应抖动消除算法和语音包丢失隐藏算法。论文首先对NetEQ的体系结构及处理流程做了概括性的分析，并重点对NetEQ的控制模块－MCU的控制机制，以及信号处理模块－DSP对语音数据的处理方法做了详细的研究和性能分析。分析结果表明NetEQ在抖动消除和丢包隐藏方面的优势。最后，本文基于语音质量角度对NetEQ做了部分优化，开发了语音质量评估模块。论文基于E-Model语音质量预测模型及ITU的单端非侵入型评估算法P.563算法进行了语音质量评估模块的设计和开发，并用UML的时序图描述了模块的开发流程及方法。该模块的评估结果基本符合主观评价。

With the rapid development of the computer network, VoIP telephone service has already been widely used. This voice communication technology which based on IP network provides a kind of convenient and cheap voice calls, and brings great changes to communication industry. But, the requirement of voice quality which appears subsequently in VoIP , is urgent.WebRTC is one of the most advanced voice engine technology, and contains the NetEQ module, which provides complicated speech signal processing. The NetEQ module provides a good solution to those problems in VoIP system, such as delay, jitter and packet loss, so widely used. But the NetEQ is a patent technology, and there is no public document to introduce or describe its detailed algorithm. Thus, a research and analysis of NetEQ is necessary for better application of it in VoIP industry.The NetEQ algorithm integrated the adaptive jitter elimination algorithm and voice packet loss concealment algorithm, which is the core contents of this paper. Firstly, the system structure and processing procedure of NetEQ have been analyzed briefly in this paper. Then, the emphasis is putted on the control mechanism of NetEQ’s control module(MCU) and the speech data processing method of processing module(DSP). The analysis results showed that the NetEQ indeed has an advantage in jitter elimination and packet lost concealment. Finally, a speech quality evaluation module, which based on the speech quality of NetEQ, is developed in this paper. The evaluation module is based on voice quality prediction model (E-Model) and single-end non-intrusive speech quality measurement(P. 563), and an UML sequence diagram to describe development processes and methods of the speech quality evaluation module is also provide in this paper. The evaluation results are mainly in agreement with the subjective feeling.