Research And Implementation Of Hearing Environment Recognition And Automatic Gain Control For Digital Hearing Aids

Hearing aid users live in a variety of environments, such as in a quiet room,noisy on the highway or at a concert and so on, like normal persons. As the single algorithms and their multitude of possible parameter settings are mostly working in a situation-specific way, that is, these algorithm are only available in certain hearing environment whereas they have no or even negative impact in other audio environment. Some hearing devices require patients to manually adjust the parameters depending on the environment and with the volume control switch to control the output volume to suit the current environment for better sound. If hearing aids devices can identify the hearing environment automatically, set up parameters and control volume automatically to adapt to the current hearing environment instead of manual operation, the intelligence of a hearing aids will be greatly enhanced.The hearing environment recognition algorithm used in this dissertation can recognize the quiet, noisy and musical environment under a variety of scenarios,such as street, train or restaurant scene in noise environment. It then adjusts the parameters of other hearing aids algorithms according to different audio environment. This dissertation presents a method to improve the training algorithm of Gaussian mixture models(GMMs), which uses fusion LBG clustering algorithm and annealing algorithm to initialize the value of GMM parameter means ?, instead of initializing it randomly in traditional way. The method can improve recognition accuracy by optimizing the GMM parameters.The experimental result demonstrates that the accuracy rate of the traditional GMMs algorithm is about 77%, while that of the method we proposed is about85%. In the end, we continue to explore the influence of the selection of the MFCC features and Gaussian mixture numbers on the recognition rate.This dissertation presents an automatic gain control algorithm(AGC) based on the average energy of speech, which keep the volume adjustment, in order to listen comfortable signal, according to the input speech signal real time to change the volume of the sound, a faint sound is amplified, and a large sound is limited.And then this dissertation presents automatic gain control algorithm based on voice activity detection. The speech activity detection algorithm is used to judge whether the current speech signal frame is a speech signal, and if the speech signal frame is enabled by AGC algorithm, otherwise it can suppress the non-speech frames.