论文部分内容阅读
为了解决目前正弦激励线性预测声码器中各子带的清浊音硬判决使得语音帧间产生明显过渡感的问题,提出一种子带清浊音模糊判决算法,直接采用自相关函数对各子带的清浊音度进行描述,矢量量化后传输至解码端,用于激励信号的产生。测试结果表明:采用子带清浊音模糊判决算法,能够消除语音帧间的过渡感,能够使各种速率正弦激励线性预测声码器的平均主观意见得分提高约0.05以上。
In order to solve the problem of the clear transitional sense between the speech frames in the current sounder excitation linear prediction vocoder, the sub-band unvoiced and unvoiced fuzzy decision algorithm is proposed. The autocorrelation function Clear voiced sound degree description, vector quantization after the transmission to the decoding side, used to stimulate the signal generation. The test results show that sub-band voiced / unvoiced fuzzy decision algorithm can eliminate the transitional feeling between speech frames and improve the average subjective opinion scores of sinusoidal stimulated linear predictive vocoder with various rates by about 0.05.