基于Ezairo DSP的ADPCM语音解码器设计  

Design of ADPCM speech coder based on Ezairo DSP

作　　者：徐晓亮[1] 梁维谦[1] 

机构地区：[1]集美大学信息工程学院,福建厦门361021

出　　处：《计算机应用》2015年第A02期319-321,共3页journal of Computer Applications

基　　金：科技部科技型中小企业创新项目(14C26213511853)

摘　　要：针对语音提示音需要占用较多Ezairo DSP存储空间的问题,实现一种8 k Hz采样和4-bit量化的自适应差分脉冲编码调制(ADPCM)解码算法,将需要解码的数据存储在DSP的EEPROM里,再经由SPI接口传输给DSP内核解码,通过简化逆量化器和索引查表规则进行算法优化,提高程序的执行效率;为适应16 k Hz的系统采样率,利用sin c插值方法将8 k Hz采样率的解码输出转换成16 k Hz系统输出。实验结果表明,ADPCM解码器能保证重建语音的质量,其MOS评分能达到4.0。在每帧2 ms的延时下,通过对代码的优化可以节省10条指令,从而提高15%的效率。Focusing on the issue that storing the voice prompts in the Ezairo DSP consumes precious memory,an Adaptive Differential Pulse Code Modulation（ ADPCM） decoding algorithm with 8 k Hz sampling and 4-bit quantization was proposed. The encoded data was stored in the EEPROM of Ezairo DSP,then transmitted to the DSP core from EEPROM via the SPI interface. Finally the algorithm was optimized so that the efficiency of the program was improved by means of simplifying inverse quantizer and indexing look-up table rules. In order to adapt to 16 k Hz sampling rate of system,sin c interpolation method was utilized to make the decoded output of 8 k Hz sampling rate converted into system output of 16 k Hz.Experimental results show that ADPCM decoder can guarantee the quality of the reconstructed speech,and the Mean Opinion Score（ MOS） can reach 4. 0,the efficiency of runing application increases by 15% though code optimizing,and 10 instructions can be saved in each frame 2 ms latency.

关 键 词：自适应差分脉冲编码调制解码 Ezairo DSP SIN c插值 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]