Research On DSP-based Voice Signal Compression Communication System

The most commonly use, the most direct, the most convenient and the most efficient way of exchanging information with people is speech, so m ost communication system has the function of speech transmission.With the development of the Internet and the concept of cloud com puting is proposed, transmission and storage of data includes video,images,text,audio and so on in the communication system of multi inform ation fusion, which need s more bandwidth and storage space.If under the premise of guaranteeing the quality of speech,system can storage and transfer speech data which is com pressed, so that can provide more bandwidth and storage space for other data and improve the utilization rate of bandwidth in system, so it is very important to research the speech compression technology.This paper focuses on research of speech encoding technology and transmission technology in communication system based on DSP, the main content includes the following several aspects:On the basis of the research about pr inciple of G.723.1 speech coding algorithm, firstly, the algorithm is implemented on VC++ 6.0. On the basis of this, the algorithm is optimized in th is paper.Through the optimization of three m odules of pitch predictor, adaptive codebook search, fixed codebook search in algorithm and function in c program, reduces the computational complexity of the algorithm.According to TCP/IP protocol, this paper uses the platform of VC++6.0, realizes the design of audio compression transmission system consists of tow PC based speech codin g algorithm in local area network comm unication system. The experim ental results show that: the com pression ratio of 5.3kbps reaches 1:22, the com pression ratio of 6.3kbps reaches 1:19 in algorithm,algorithm has good compression effect, the voice is clear and can be understand, MOS points of quality of voice is good.On the basis of realization of G.723.1 speech coding algorithm, this paper realizes the transplantation of algorith m on DSP processor. Through the use of intrinsics, special instruction other a ssembly optimization methods,this paper implements algorithm on DSP processor in r eal-time. The results after optimization showes that: occupied clock cycle s are re duced by at least 90% than before after optimization in partial function.T o a frame of voice,encoding and decoding time of algorithm at 5.3kbps is reduce by 68% than before, algorithm at 6.3kbps is reduce by 64% than before, MOS points of quality of voice is good.Finally, this paper com pletes the design of the speech co mpression communication system between DSP and PC. This system realizes the Ethe rnet communication between DSP and PC, the acquisition of the speech signal,encoding, sends speech data based on UDP protocol, and realizes the functions of receivin g audio data,playing sound after decoding, veri fies the system can transfer speech in real-time, and quality ofspeech is good, sati sfies requirement of the quality of communication.