由嵌入式设备所构建的视频会议系统终端,在会话建立后,需要对来自各方的媒体信息诸如语音进行混音处理。在信令集中、媒体流分布的系统中,媒体的混合没有独立出来放在Focus中,而是将与会的各终端发来的语音媒体流在会议参与者主机上实现端混合。在众多混音方案的比对中,这里介绍一种改进的符合语音信号特征的混音架构及溢出处理算法,并用语音信号的短时能量及短时过零率对混音过程进行动态修正。在嵌入式系统的实现验证中表明,该算法提高了混音溢出处理的效率,降低了噪音,具有较低的算法复杂度,并有良好的听觉舒适感。 Video conference system terminals built by embedded devices need to mix media information such as voice from all parties after the session is established. In the system of signaling concentration and media flow distribution, the mixing of media is not independent of Focus. Instead, the voice media streams sent by each terminal participating in the conference are mixed in the host of the conference participants. In the comparison of many mixing schemes, here is an improved mix structure and overflow processing algorithm that is in line with the characteristics of speech signals, and the mixing process is dynamically modified with the short-time energy of the speech signal and the short-time zero-crossing rate. The verification of the embedded system shows that the algorithm improves the efficiency of the mixing overflow processing, reduces the noise, has a lower complexity of the algorithm, and has a good sense of hearing comfort.