Research On Noise Cancellation Algorithms For Vehicle Hand-free System And Hardware Implementation

Noise and echo inevitably exit in the voice communication of vehicle hand-free system,which affect the voice quality in the real-time audio communication.With the development of technology,the theorys of noise controlling have made a great progress.In the experimental phase noise the effect of noise cancellation is obvious by adaptive filtering algorithms,which are often used in practical applications.However,they are restricted by the hardware platform resources,processing speed and price.Therefore it is very important to seek a filtering algorithm with low computational complexity,good stability and fast speed of convergence.The Least Mean Square(LMS)algorithm is a popular adaptive filtering algorithm for its simple structure and low computational complexity.But its speed of convergence is very slow.The step can be fixed to solve this shortcoming.So many variable step size LMS algorithms have been improved.To reduce the complexity an improved algorithm is proposed by bit computing to update filter coefficient.The experimental results show that the algorithm has lower computational complexity with simlar convergence speed.A step control matrix is introduced in Proportionate Normalized Least Mean Square(PNLMS)algorithm.This considerably improves the initial convergence speed by setting different steps for different filter coefficients.However,the later convergence speed becomes slower.To solve this problem,an improved PNLMS algorithm is proposed.A nonlinear reciprocal relationship between proportionate step and current filter coefficients' amplitudes is established by quantitatively analyzing of the convergence process.The simulation shows that the improved PNLMS algorithm has faster convergence speed and better robustness than PNLMS and MPNLMS algorithms.Moreover,the computational complexity of the algorithm is reduced a lot,even much lower than the MPNLMS algorithm.The noise and echo should be cancelled in bluetooth system.In this dissertation,the adaptive filtering algorithm is realized and transplated into the STM32 platform.Experimental results demonstrate the efficacy of the proposed adaptive filtering algorithms.