| Summary: | 碩士 === 世新大學 === 資訊管理學研究所(含碩專班) === 101 === With the advent of the global village, good-sized enterprises and organizations
often use multi-point video conferencing to have remote large video conferences and educational training. Although video conferencing is certainly demanding for images, smooth and clear voice is also one of the key points to make meetings run smoothly. Because the sound passes in the space, the voice from the speaker will be recovered by the microphone again. This causes the generation of acoustic echo. If a member at the conference produces an echo during the meeting, this echo will affect all the members who are participating in the conference. It will lead to a serious decline in the quality of the conference. Therefore, the function of acoustic echo cancellation
must be necessary.
Owing to the continuous upgrades of video quality, the CPU of the Video points
must handle large compression, decompression rates of the image. Consequently, the efficiency of echo cancellation algorithms needs to be improved, so that it won't affect the quality of video conferencing. Besides, efficiency improvement must proceed without affecting the effectiveness of the conference.
This thesis aims to study the software technology on acoustic echo cancellation.
This study proposes to use the EAU-MDF algorithm to improve the efficiency of the MDF algorithm. It also develops a multi-point video conferencing system to do field testing experiments. The results show that the EAU-MDF algorithm has nearly 45 percent more efficiency than the MDF algorithm.
|
|---|
