Adaptive Speech Enhancement In The Battlefield Environment

This paper carries out a new method of Adaptive Filters and Singular Value Decomposition (SVD), which can effectively improve the quality of voice signal communication in battle field by eliminating the wide-band and random background noise.Firstly, the paper proposes a speech enhancement system with two-channel input signals. And through analyzing the generation and spread paths of the ambient noise, a simulation model is set up.Secondly, according to the relativity of the signal, we have studied the similarities and differences between ambient noise and speech signal in battle field, and put forward a block diagram of signal enhancement system with the character of time delay. Simulation result proves the system is available. Furthermore, the loss of the original signal is also discussed in detail.Then the paper analyzes the relation between two important factors of SVD, and improve the formula of singular entropy increment to make it can get the best filtering result under various kinds of SNR.The hardware design of the system based on DSP is introduced at the last part of the paper. Experiment results show the system is stable and the real-time character is satisfied.