Speech Enhancement Technology In The Acoustic Feedback System

Speech signal is not pure, but often corrupted by different kinds of random noise which disturbs the processing quality of speech signal. Therefore, speech enhancement technique plays an important roll in speech signal processing.Speech enhancement technique is to make the speech signal corrupted by noise, more clear, understandable and comfortable. Most practical speech enhancement system composed of an accuracy noise estimator and a good filter to reduce the noise.The noise estimation includes VAD and updating noise spectrum continuously. The first one is complemented easily, the later aims at actual unstable noise and updates the noise spectrum continuously. How to use these two methods is the key of speech enhancement algorithm. If the computation complexity is permitted, we can combine two methods together, which is the improvement trend of noise estimation.There are many speech enhancement techniques, such as filtering, correlation of speech, nonlinear processing, spectrum subtraction, and the adaptive Wiener filter. The research content of the thesis is the speech enhancement technique that is used in the acoustic feedback suppresser.Firstly, we searched and compared the methods of noise estimation based on VAD and updating the noise spectrum continuously, combined them together to make some improvement.Secondly, we research on some speech enhance techniques including short time spectrum analysis speech enhance technique and its improvement form, simulated the algorithms and compared them each other.Thirdly, based on searching the noise estimation methods and speech enhancement techniques, we designed a speech enhancement system and simulated the algorithms on TMS320VC5510DSK.