The Research On Microphone Array Speech Enhancement Algorithms In Digital Hearing Aids

Digital hearing aids can effectively ameliorate the aural comprehension of the patients.However in practical enviroment, reverberation, resound and other various interference sources caninfluence the performance of hearing aids, the introduction of microphone array can improve theseproblems, attain effective speech enhancement. In this paper, the key technologies in digital hearingaids are researched, the microphone array speech enhancement is mainly studied, then this paperproposes the improvement and optimization. The main work of the paper is as follows:Firstly, this paper researches the working principle of digital hearing aids and the basic theory ofspeech signal, analyzes auditory mechanism. The common single-microphone speech enhancementalgorithms and other key technologies, such as sound source localization, frequency transpositioncompression, loudness compensation and feedback cancellation are researched and thecharacteristics of each technology are analyzed.Secondly, this paper in-depth researches microphone array speech enhancement algorithms,analyzes the principles and performances of delay-sum beamforming method、GSC method、Zelinski method. For the shortage of traditional delay-sum beamforming method that it can hardlyeliminate coherent noise, the improved method is proposed by combining effectively with thewavelet threshold denoising, to improve the accuracy of time delay estimation, and the model afterimproved can not only eliminate incoherent noise but also eliminate the coherent noise.Finally, this paper makes a theory analysis on the speech leakage and residual noise ofgeneralized sidelobe canceller (GSC) adaptive beamforming algorithm, discusses the characteristicof speech leakage caused by the direction estimation error. On the basis of the analysis andresearch, proposes a speech compensation method. Via using voice activity detection (VAD) differspeech frame and noise frame for the result of GSC, then make a enhancement for speech frame andmake a reduction for noise frame, to achieve a effective compensation and reduce residual noise,the simulation and test results show that the compensation method can further improve the effect ofspeech and improve the rate of speech recognition.