Research And Implementation On Speech Enhancement Of Shortwave Radio

Due to the transmission distance ,the simple equipment and mature technology ,short-wave radio communicate have been widely used in the military and civilian.But its channel complex and highly suspectible to nature effects and human inteferrence serious impact the communication quality.Therefore,how to effectively reduce background noise,accurate detection and filter out a variety of narrow-band interference has always been a hot research and has a high practical value.Based on in-depth study of shortwave voise signals ,we establishd the background noise and narrowband interference hybrid model and designed the short-wave voice denoising model. The mode is tested and is found to perform effectively.The main contents and achievements are as follows:1,A novel voice activity detection algrotithm has been proposed based on wavelet entropy according to the characteristice of shore-wave speech signals.First , a wavelet transform is used to decompose the resulting g speech signal into critical bands.Then according to the energy entropy differcece between speech frame and noise frame in wavelet domain,they have been distinguished under the application of Hangover process.The new algorithm is tested over a low SNR conditions and is found to perform better than traditional ways.2,After analyzes the residual musical noise issue of statistical model speech enhancement algorithm for MMSE spectral amplitude estimation and MMSE-LSA estimated at low SNR condition.,A new speech enhancement with enengy entropy speech modeling was described. Simulation results show that this algorithm yields less music noise .3,For the narrow band interference problems in short wave speech,a new metherd which employs IMCRA (improved minima controlled recursive averaging algorithm)Noise estimation algorithm and traditional speech interference detection algorithm was proposed.First,voicing voice signal according to the distribution of the characteristics of ench frame signal in the frequency domain in sub-band.Then,interference threshold was calculate according to the mean and various points of the sub-band energy.The algorithm is high precision and effectively protect the voice frame which has high-energy harmonic components.4,When the narrow band interference and voice signal spectrum overlaped, the traditional interference rejection algorithm would cause severe financial losses.This paper presents a norval methord which combin traditional narrow-band interference suppression algorithm to adaptive notch adjust algorithm. It effectively improved the SNR and reduce the distortion which achieved good hearing results.5,By the test results using the recursive iteration method on the current frame noise estimation and statistical speech enhancement models,this algroithem effectively reducing the background noise and narrow band interference to improve the SNR and speech intelligibility.