Research On Short Wave Voice Enhancement And Combining Technology

The frequency range actually used by short-wave communication is 1.5MHz~30MHz.It mainly uses ionospheric reflection to transmit information,which is a more effective communication method.For many years,short-wave communication has been used in military communications and other fields due to its high reliability,good security and maneuverability.Voice signals are the main service of short-wave communication.When the voice signal passes through a short-wave channel,it is inevitably contaminated by various noises and interferences,so that the received signal is no longer pure voice.One of the key technologies for processing noisy voice signals to restore pure voice signals as much as possible is single-channel voice enhancement.However,the effect of single-channel voice enhancement can not meet the actual needs under the short-wave channel,so a new voice enhancement technology is needed.One of the main features of the short-wave channel is multipath effect.When the voice signal passes through the short-wave channel,multipath fading occurs.Currently,one of the the effective techniques for resisting the multipath fading is the diversity combining technique.The diversity combining technique is mainly used for digital communication.Based on the single channel voice enhancement,this paper introduces the diversity combining technology to merge and receive the multi-channel voice signals,and it makes full use of the diversity combining gain of the multi-channel voice signals,which further improves the voice quality.This paper mainly deals with multi-channel receiving short-wave voice signals.The key technologies include single-channel voice enhancement technology,multi-channel voice synchronization technology and multi-channel voice signal diversity reception and combining technology.Aiming at the single-channel voice enhancement technology,this paper proposes a Spectral Subtraction Based on Masking Threshold for Multi-Window Spectral Estimation and Adaptive Noise Estimation(SSBMT-MWSE-ANS).The algorithm uses multi-window spectrum estimation combined with adaptive noise estimation algorithm to adaptively estimate the single-channel voice signal and noise power spectrum,and the speech signal is denoised by spectral subtraction,so that the quality of the enhanced speech signal is improved.Aiming at the multi-channel voice synchronization technology,this paper uses the frequency domain matching correlation-based delay estimation algorithm to estimate the delay of multi-channel voice signal.The algorithm needs to process the speech by segmentation.Therefore,this paper proposes an energy-entropy ratio endpoint detection algorithm based on interference detection.The algorithm performs endpoint detection on the voice signal with interference,and it can accurately obtain the start and end points of the voice.In this paper,the phase correlation algorithm based on cross-correlation is used to estimate the phase difference of multi-channel voice signals.The results show that these two algorithms accurately estimate the delay and phase difference of the multi-channel voice signals under the short-wave channel.Aiming at the diversity receiving and combining technology of multi-channel voice signals,this paper proposes a weighted combining algorithm based on the P.563 scoring algorithm,which uses the P.563 scoring result as a weighting coefficient for weight combining in the speech segment,and equal gain combining in the silent segment,so as to improve the quality of the voice signal.Through these technologies,various noises and interferences existing in wireless short-wave voice can be effectively suppressed or eliminated,and the intelligibility and comfort of the voice signal can be improved.This paper proves that the weighted combining algorithm based on P.563 scoring algorithm can effectively merge the simulated data and the actual collected data by scoring the comparison results.