| The shortage of terrestrial resources urges the development and utilization of Marine resources.Therefore underwater acoustic communication is getting more and more important.Since the complexity of signal path in water, media and bandwidth resources are extremely limited.It makes sense that people are trying to compress voice signals and downgrade the number of bits from signal sources.MELP is a voice compression coding language which has excellent compressibility and can also encrypts the signals.Based on underwater acoustic communication technologies, this thesis analyzed the features of underwater acoustic communication.Then by applying MELP,this thesis simulated the process of low rate under water voice signal compression on Matlab and transplanted the algorithm to DSP6713 successfully.In the first part of this thesis,this thesis illustrated the MELP coding algorithm with details which including calculation and quantification of pitch period and amplifier,analysis and quantification of linear prediction coefficients, adaptive spectral enhancement, synthesis of linear prediction, adjustment of amplifier, pulse shaping filter, etc. The result of the MATLAB simulation verifies MELP speech coding features. Then a brief introduction of DSP6713 chip is given. According to the MELP speech coding requirements of real-time, the hardware platform is built which mainly includes configuration of DSP/BIOS, EDMA and McBSP, distribution of memory spaces,the configuration and adjustment of AIC23 which is the most important part.After the illustration of DSP development environment, this paper studied the data-structure and implementary method of MELP algorithm in C Language.Then the MELP compression algorithm was transplanted to DSK6713 platform. Furthermore, the program was optimized to reduce time-taken and enhance efficiency.At the end of this thesis, the software design process of this algorithm in real-time DSP platform is described. Then make a detailed optimization and debugging to improve program.The single-frame and multi-frame speech signal were used to do some tests where the consistency and understandability of MELP Coding/Decoding algorithm were verified by comparing with the original voice signals.Finally,this thesis make a subjective test on the performance of the algorithm. |
