Study On Microphone Array Speech Enhancement Algorithm

As a natural way of interaction between human beings, speech communication plays an important role in people's daily life. However, speech is inevitably corrupted by ambient noise in the real environment. The existence of background noise can affect the listener's perception and turn down the performance of many speech processing systems. Speech enhancement solutions are expected to extract desired speech from noisy speech signals, improve quality of speech and offer comfortable listening. As a result, speech enhancement has become indispensable in speech communication.Microphone array speech enhancement is the most common method of speech enhancement. Based on the existing microphone array speech enhancement algorithms, an improved algorithm is studied in the thesis. The classic Generalized Sidelobe Canceller structure has a disadvantage of signal leakage, in other words, the noise reference signals contain the desired speech which leads to cancellation of the desired speech. In view of this shortcoming, an improved subband adaptive beamforming algorithm combined with subspace noise reduction algorithm is studied. The algorithm improves the adaptive algorithm in Generalized Sidelobe Canceller structure which degrades the affect of signal leakage successfully. Subspace noise reduction algorithm is followed to remove residual noise. The simulation results show that: compared with traditional subband adaptive beamforming algorithm, the improved method has better output signal to noise ratio and the quality of speech is enhanced obviously.