The Research And Design On Speech Enhancement System Based On DSP

With the rapid development of science and technology, voice communications technology is becoming more and more mature. Voice is a very important way of communication between humans, and voice quality will affect the communication and exchange between people. In real life, when using mobile phone, microphone, telephone and other communications equipment in such as military exercise site, the concert scene, and vehicle communication the quality of voice will be inevitable affected by the noise from the environment and make the quality decline, seriously will lead to the conversion impossible. Therefore, communication in strong background noise, it is necessary to filter out the background noise. The main purpose of speech enhancement is to remove background noise and improve voice communication system performance in the noisy environment.In recent years, with the development of DSP(Digital Signal Processingy) and VLSI (Very Large Scale Integrated Circuit), providing a theoretical and technical supporting in theory and technology. With the continuous improvement and development of DSP and theimprovement of developing platform, enabling the voice enhancement technology implemented well in the hardware system, and can be well applied in different occasions of voice communication.In this paper,the vioce enhancement in noisy environment is studied,the mainly purpose is to improved the quality of voice communication, as far as possible to improve the intelligibility and clarity of voice in communication, creating a comfortable environment for voice communication. Voice Activity Detector(VAD) as the front processing algorithm of the speech enhancement,it's performance directly affect the speech enhancement algorithm。In this paper several VADs are studied, and also simulate these algorithms,The results are analyzed and compared, ultimately the VAD based on frequency deviation as the voice activity detection of the system.The main link of the speech enhancement is the achievement of the speech enhancement algorithm. The spectral subtraction becomes the important method for speech enhancement because of the simplicity and effectivness, but the traditional spectral subtraction often associates with "muical noise", at the same time the speech distortion is relatively large after speech enhancement. Especially when the SNR is relatively low, the performances of these algorithms are sharply decline. The spectral subtraction is optimized and improved based on the analysi and study of several speech enhancement algorithms. Avoice activity detection based on multi-band spectral subtraction speech enhancement algorithm is used. In this paper, we test and simulate the algorithm with the background noise of different signal to noise ratio. The results show that the algorithm can effectively inhibit the "music noise", at the same time, increasing the signal to noise ratio and improving the quality of voice communication in the noisy environment.The article introduces the design of hardware and software based on TMS320C5509 DSP, also the connections between system peripheral devices and DSP and Initial configuration of each module and algorithm implementation process on the system is introduced. At last the improved speech enhancement algorithm is successful operated on the hardware system. The system can play an inhibitory effect on the background noise, the basic purpose of speech enhancement is achieved.