Implement of MEMS Microphone Array for Robust Speech Recognition

• Main Authors: Hung-Yi Chang, 張鴻毅
• Other Authors: Yuan-Fu Liao
• Format: Others
• Language: zh-TW
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/b2d4nj

碩士 === 國立臺北科技大學 === 電腦與通訊研究所 === 96 === As the demand for voice technology, the microphone array applications become more diverse. For example, in the handheld device noise and echo suppression and audio tracking and speaker recognition, etc.. In this paper we design a microphone array for the simultaneous capture information and the establishment of a complete system architecture and the test of various algorithms. Therefore, we use a MEMS-based microphone array. The MEMS manufacturing process easier and chip combine to low power consumption, small size and other advantages. We implement the microphone array and a variety of algorithms for processing, confirmed this system function and action correct. We implement sound source tracking, Delay-and-Sum beamformer, adaptive filter beamformer and Subband Maximum Likelihook Beamformer. In sound source tracking application, we have an excellent track results. In speech recognition algorithm validation, and the results are not as expected. After analysis, the circuit is making noise reduced SNR, thus affecting the outcome of recognition. In the future we will improve the circuit performance. The microphone array can be made to maximize the effectiveness of achieving the use of the microphone array.