【摘要】 语音编解码算法是数字对讲机的核心。由于数字对讲机信道带宽和信道质量的限制,要求采用的数字语音编码算法在保证语音质量的情况下降低编解码速率,并且要求语音编解码算法有一定的容错能力。由于数字对讲机手持设备功耗的限制,要求数字语音编解码的算法复杂度尽可能低。文章对正弦激励线性预测(SELP)的2.4k语音编解码算法进行简化,在基本不降低语音质量的条件下大大降低了算法复杂度,并针对ARM9E硬件平台进行优化,最终在该平台上实现了该算法。更多还原

【Abstract】 Voice encoding and decoding algorithm is the core part of interphone system. Due to the restriction of bandwidth and quality of the telecommunication channel,the encoding algorithm should keep the bit-rate low while maintaining the quality of speech signal and tolerating a certain degree of faults. And the limited power consumption of handset requires that the complexity of the encoding algorithm be confined in a reasonable level. In this paper,the 2.4k bit Sinusoid Excited Linear Prediction(SELP) speech encoding algorithm is improved and simplified,and the complexity of the algorithm is greatly reduced while the quality of the speech signal is well maintained. In the mean time,the object code is optimized and the algorithm is implemented on the ARM9E processor.更多还原