Research Of Echo Cancellation Algorithm For Wavelet Domain

With the progress of era, the development of science and technology brings business opportunities to each telecommunication equipment and also poses challenges. Along with the increase in communication equipment type, user’s voice in the voice equipment is becoming more and more high quality requirements. It is a basic condition that the communication device can have a clear voice for voice communication products.However, because of complicated living environment, most of the speech signals are subject to different levels and different types of the echo signal interference. The qualities of voice calls during the process of calls are affected. It is particularly important to echo cancellation effectively for communicate equipment.(1) Based on the research of the common methods of acoustic echo cancellation, this paper presents improvements of adaptive echo cancellation module and double side detection module in the system of adaptive echo cancellation, respectively. The research of double side calls detection algorithm. The article explains many common detection algorithms. Enumerates the double end detection algorithm is common, due to the shortage of the enumeration algorithm, an improved wavelet domain more eigenvalues double-end detection algorithm on the basis of the traditional double end detection algorithm is proposed in this paper. The improved algorithm FIRstly wavelet decomposition the input signal and reference signal of the adaptive filter, and wavelet threshold denoises the signal wavelet subband decomposed. Then make a double threshold energy a judgment on the low frequency wavelet subband denoised in different states, whether mute state, active state or call state. When the speech signal is in mute state or call state, the adaptive filter could make corresponding process according to its decision. While in active state, the signal becomes less due to the increase of the energy, it cannot be decided the state by once time. Then we should compare the gene cycle of speech signal in active state and judge the state of active state in call so that the adaptive filter could make the corresponding process.(2) For adaptive echo cancellation algorithm, this paper FIRst enumerates the common adaptive filtering algorithm, This paper FIRst lists the common adaptive filtering algorithm, Due to the shortage of the enumeration algorithm, this paper puts forward an improved adaptive hybrid conjugate gradient iteration algorithm. This algorithm combines mean algorithm combined with hybrid conjugate gradient iteration algorithm. It is used to the adaptive filter weights updated.This paper shows simulation experiments about the presented wavelet domain multiple eigenvalues double-end detection algorithm and the improved adaptive hybrid conjugate gradient iteration algorithm, respectively. The experimental results substantiate that the improved algorithm has good performance, and consistent well with theory.