Microphone Array Speech Enhancement Based on Filterbanks Generalized Sidelobe Canceller
|School||Guilin University of Electronic Science and Technology|
|Course||Signal and Information Processing|
|Keywords||Speech Enhancement Microphone array Filter bank Sub-band technology Endpoint detection Generalized Sidelobe Canceller|
Speech signal processing is the study of a subject of voice signal processing using digital signal processing techniques. The voice signal processing is extensively used, one of the main technologies including speech coding, speech synthesis, speech recognition and speech enhancement. The practical application of the environment due to the presence of background noise, seriously affecting the application of these technologies. How to effectively remove various types of noise, it becomes very important to get useful speech information. Speech enhancement technology is gradually moving towards the practical with the times, has appeared in a variety of enhancement algorithms. This paper first introduces the currently widely used several speech enhancement algorithm, and briefly discusses the advantages and disadvantages of each. Traditional speech enhancement based on odd and even a microphone, in this case, speech and noise at the same time present in the one channel, the voice information and noise information must be derived from the same signal. In a multi-channel voice system, each channel there are some characteristics, these characteristics of speech enhancement processing is very favorable, and the microphone array to take advantage of the spatial information, thereby microphone array speech enhancement has become a hot research topic. Disciplines of the signal processing, the sub-band speech enhancement techniques is the signal through the different bands of the filter, the signal is decomposed into sub-band of the different frequency bands, so that each sub-band signals may be respectively the corresponding processing, and reduce processing time. Perfect reconstruction filter group to be widely used in a sub-band coding, but these filters group is less suitable for the sub-band adaptive filter, because when the sub-band signals by having an arbitrary amplitude and phase response of the filter, and these are completely re The configuration filter banks eliminate aliasing affect performance can not be guaranteed, it is studied in this paper a subband adaptive filter bank. In this paper, sub with a combination of technology and microphone array theory, design a filter bank generalized sidelobe phase canceller (GSC) microphone array speech enhancement structure, the structure with a classic voice endpoint detection technology, namely the sub-band signal in judging voice segment noise segment, the adaptive filter coefficients, respectively, to make a deal. In-depth theoretical analysis on the basis of, respectively, based on filter bank GSC microphone array speech enhancement system under different noise environment simulation, simulation results show that, to a certain extent based on filter bank GSC speech enhancement system elimination The environmental noise, but there are still some defects, for these drawbacks, the authors proposed some measures for improving the structure. The simulation results show that the improved system scattering noise and extraneous noise environment still has good noise cancellation and anti-jamming performance, worthy of further research. Finally, a summary of the results and conclusions of this article, the few comments about the future research work.