Design Of MELP-Based Speech Compression Algorithm With Different Rates And Its DSP Implementation

With the gradual deepening of China’s ocean development,underwater environmental monitoring and underwater resource development play an important role in people’s production and life.With the diversification of communication methods,underwater acoustic communication has gradually become well-known to the public.However,due to the late start of underwater acoustic communication technology,the technology is not yet mature enough,and the complexity and variability of underwater acoustic channels,spatiotemporal frequency changes and multipath effects have become difficult problems in underwater acoustic communication research.The current underwater communication technology mainly includes wired communication,acoustic communication,laser communication,and radio communication.These technical methods have their own advantages and disadvantages,but underwater acoustic communication is the most reliable.Therefore,in the underwater environment with little channel capacity,the technology of continuously optimizing the transmission rate will occupy a higher proportion in the future research.Firstly,a study is conducted on the traditional Mixed Excitation Linear Prediction(MELP)speech coding algorithm.By analyzing the performance of different parameters in speech coding,the MELP speech coding algorithm is selected to improve its speed reduction.Then,based on the 2.4kbps MELP speech coding algorithm,the speed of the MELP speech coding algorithm is improved.The improvement of rate is mainly to choose speech coding algorithms with different rates to adapt to different channels.Subsequently,this article studied MELP speech coding algorithms with variable bit rates of 2.0kbps,1.5-1.8kbps,0.8kbps,and 0.6Kbps,respectively.The 0.6kbps MELP speech coding algorithm mainly uses multi frame joint quantization technology to reduce the rate.The principle of 0.6kbps is to form a super frame of three subframes,and then jointly quantify their parameters.According to the Line Spectral Frequency(LSF)parameter,by quantifying it at 24 bit and quantifying the gain at 10 bit,we chose to increase the time of one frame from 22.5ms to 30 ms.Through the above method,we achieved a MELP speech coding algorithm of 0.6kbps.After objective speech testing and scoring,the scores of PESQ and STOI were also in a relatively ideal range.At the same time,it is necessary to create a corresponding upper computer control system for the lower computer.Finally,implement a 0.6kbps MELP speech encoding algorithm and upper computer control system on hardware using the above techniques.In order to address the issues of low communication rate and cost reduction in underwater voice communication,this paper conducts research and implementation of speech coding algorithms with different rates.The algorithm implementation platform selected a DSP platform that uses low-power and cost-effective TMS320VC5509 A for algorithm transplantation,and optimized the algorithm.Finally,the reliability of the MELP algorithm with different bit rates for underwater communication systems was verified through experiments.The upper computer software of the system has completed the control of the lower computer and the corresponding data acquisition function,and the entire system has great development value.