| At present, military radio of our arms suffers from the environmental noise. The speech quality of the system degrades especially in the presence of heavy noise. Therefore, how to obtain the original speech from the noisy signal is the main problem to be solved in the field of military communication.Speech enhancement is an important research direction of the speech signal processing, the main objective of which is to extract the original speech signal as pure as possible from the corrupted speech signal and thus improve the aural quality and enhancing the speech intelligibility. According to the difference of processing, speech enhancement methods can be divided into four classes: parametric, nonparametric, statistic and other methods.Based on the both noise cancellation and spectral subtraction algorithms, a modified speech enhancement method for the military radio is introduced in the paper. In addition, the proposed front-end speech enhancement system for the military system is also designed and implemented in the hardware.The outline of the paper is organized as follows:(1) Briefly summarizing the researchful meaning, applied background and study development status of home and abroad in speech enhancement field, studying some current conventional methods and analyzing and comparing to their performance.(2) Considering actual application, properly improving noise cancellation method and combining spectral subtraction method, an improved combined speech enhancement method based on double microphones is proposed. Numerous simulation and actual test results show that the method is feasible and valid.(3) The improved method is implemented on TI TMS320C55xDSP. On the TMS320VC5509 DSK developmental platform of TI Company, we design the speech sampling channel of double microphones and convert channel of A/D and D/A. A front-end speech enhancement processing system of military radio is implemented and the interface test of the system with partial military radio is completed. |
PDF Full Text Request