| Summary: | 碩士 === 國立中山大學 === 資訊工程學系研究所 === 103 === Though there are many types of meeting room microphone control system in the market, the most common control method can be classified into either push-button or voice volume. Voice-activated microphones can automatically turn on the device based on the input volume. However, in a noisy environment, an error condition might occur, and it would cause further problems like error audio broadcasting and echo. Those factors would seriously affect the quality of the conference. In view of this, we proposed a multimodal microphone control system to solve those problems. In addition to using the microphone volume, we analyze the behavior of the speaker through a miniature camera. Once a speaker''s mouth moves or he does a specific behavior, the system will automatically transform the microphone control system status based on the behaviors. Through intensive experiments, the results prove that the proposed system can response to the behavior of the speaker immediately.
|
|---|
