用DSP TMS320VC5416实现G.723.1编解码算法

为了能在低码率下达到较高的合成语音质量，对双速率语音编解码算法G．723．1进行研究，并对该算法的高编码速率（6．3kb／s）的源代码进行优化，改卷积运算为加减运算，改乘法运算为减法运算，大大减小了运算量，提高了运算速度；解决了算法移植的问题，即找到语音算法与开发板语音程序的切合点——将管道传输函数插入语音程序。研究结果表明，该实时系统在6．3kb／s有较好的语音回放效果，除音量略有降低外，语音信号的失真度在人耳分辨范围之外；语音波形从时域变换到频域，解码后波形相对原波形幅值降低，波形保持不变。

Realization of algorithm of G.723.1 with DSP TMS320VC5416

In order to get high-quality compound speech signal,the algorithm of G.723.1,encoding and decoding algorithm,was discussed and its source code was optimized.By changing the operation from cirrus to adding and from multiplication to subtraction,the quantity of operation was reduced and its speed was advanced.The algorithm was transplanted to digital signal processing successfully because the key technology was worked out,that the pipeline data transmission function was inserted to main function.The results of the experiment show the real-time system can ensure good voice replay quality at 6.3 kb/s coding speed: ears cannot distinguish the difference of voice except that the volume falls.When speech signals are transferred from time domain to frequent domain,the amplitude of decoded wave is lower than that of encoded wave and the form of waves remains the same.