The Research On Time Delay Estimation And Speech Enhancement In Digital Hearing Aids

Digital hearing aids can effectively help hearing impaired patients improve their listening comprehension. The performance of digital hearing aids depends on the key technology, such as speech enhancement and loudness compensation. This paper studied the key technology of digital hearing aids, and focused on the research of time delay estimation and speech enhancement technology, the main work can be summarized as follows:Firstly, this thesis introduced the basic knowledge of speech, including feature and mathematical model. And then, studied the topology and direction chart of microphone array, analyzed the working principle and key technologies of digital hearing aids, such as sound source localization, speech enhancement, acoustic feedback cancellation and loudness compensation.Secondly, this thesis delved into time delay estimation algorithms, among conventional algorithms, least mean square(LMS) algorithm and recursive least square(RLS) algorithm are the most widely used. LMS algorithm has slow convergence speed and high misalignment, RLS algorithm has large computing and high algorithm complexity. In response to these shortcomings Combine LMS algorithm and RLS algorithm, a new time delay estimation algorithm was proposed. New algorithm performances well in convergence speed, misalignment and algorithm complexity.Thirdly, a speech enhancement algorithm based on generalized sidelobe canceller(GSC) was proposed. This paper focused on microphone array speech enhancement algorithms, analyzed the principles and performances of fixed beamforming method, GSC method. Conventional GSC adaptive beamforming method has direction estimation error, so it inevitably makes speech leakage. For this shortage, improved method is proposed by using the correlation of signal after conventional GSC processing and signal after blocking matrix. The proposed algorithm can newly estimate the direction of target speech signal, then reduce speech leakage. Simulation results show that GSC proposed method reduces the loss of original speech signal and improves comfort of listeners.