Speech Enhancement In Conference Phone Based On DSP

New technology is moving towards freedom and concealment. Hands free systems have become a popular form of communication, due to the lack of restrictions they allow. High-quality sound capture is the most important in hands free voice interface application systems. The acoustic echo cancellation (AEC) technique and Speech Enhancement technique have become hotspots in the communication world.This thesis represents acoustic echo cancellation in voice communication system. AEC is implemented by adaptive filter. In this paper some adaptive filter algorithms are discussed. They are Least Mean Squares(LMS), Normalized LMS(NLMS) and Recursive Least Square(RLS) algorithms. Their performances of acoustic echo canceller is evaluated. At the same time, Double-talk detector is used in acoustic echo canceller. In order to improve the speed of calculation, block frequency-domain LMS algorithms is represented.This thesis represents speech enhancement technique based on microphone array. A microphone array is a system that consists of multiple microphones or loudspeakers placed at the different location. Speech enhancement based on microphone arrays m ainly relies on beamfomring. Generally, arrays connect with beamformerth at provides multiple styles of special filtering. This paper mainly discusses several typical algorithms in detail, such as, delay-sum beamforming, constant directivity beamforming, generalized sidelobe canceller, and Wiener post-filtering beamforming. The simulation and analysis of these algorithms are represented.The work and innovation in the thesis:1. Comparing and analyzing a few adaptive filter algorithms and double-talk detection algorithms.2. Realizing acoustic echo canceller algorithm with block frequency domain LMS algorithm and double-threshold talk detection algorithm.3. Developing array technology in voice enhancement.4. Developing and comparing the generalized sidelobe canceller algorithm and adaptive post-filtering algorithm for noise reduction...