| Microphone array speech enhancement has become one of the hot research topic inrecent years. As the rapid development of electronic communication products, there is anurgent need to modify the quality of speech communication. The microphone array speechenhancement, is widely used in a variety of speakers, TV, teleconference and some militaryproducts.Firstly, this paper discribes the basic knowledge of microphone array speechenhancement, and then LMS algorithm adopted in the multi-channel speech enhancementsystem’s study. On the basis of studying and summarizing former researches, the authorproposed an speech enhancement method based on improved dynamic step factor LMS. Thespeech enhancement algorithm was choosen mainly aims to fit the actual bearing capacity of DSP hardwaresystem’s design. This method is simple and easy for hardware realization. By MATLABsimulation, the experiment results show that it has faster convergence speed and smallersteady-state error. Subjective and objective evaluation are used to analysis enhancement effect,output voice has better auditory perception.Then, the paper gives the system design of the four channel DSP speech enhancementbased on TMS320VC5509A. Focusing on the design content and process, including hardwaredesign and software program. The hardware design work contain the DSP core circuit chosen,speech acquisition array structure design, filter circuit and etc. The software design partincludes the design of software process and algorithm C program code, initializes eachcommunication module, expand external storage space and bootloader.Finally, in the actual acoustic environments, some experiments are done to demonstratethe effectiveness of the proposed system. The experimentl results show that the design ofmulti-channel speech enhancement system can meet the requirements of real-time speechsignal acquisition and enhancement. In addition, some advantages and shortcomings of thepaper are also given in the end of the paper. |
PDF Full Text Request