摘要
共轭代数码本激励线性预测(CS-ACELP)语音编码算法在8 kb/s速率上获得了比较理想的质量,是以10 ms为一短时语音帧作为处理对象.基于CS-ACELP语音编码算法,以20 ms为一语音帧,在编码器中引入脉冲散布技术,提出了一种使码速降低至4 kb/s的散布脉冲代数码本激励线性预测(PD-ACELP)编码算法.经仿真实验及主观听觉测试表明,这种算法的合成语音质量还是比较令人满意的.
The quality of the conjugate-structure algebraic-code-excited linear-prediction (CS-ACELP) algorithm is fairly high at the bit-rate of 8 kb/s, which employs a 10 ms -short-time voice-frame as a processing object. On the basis of CS-ACELP algorithm, taking 20 ms-short-time as a voice-frame and using pulse-dispersion technique, a PD-ACELP algorithm is proposed that decreases the bit-rate to 4 kb/s. Both computer simulations and subjective hearing tests show that the synthetic speech quality of the coding algorithm is fairly satisfactory.
出处
《应用科技》
CAS
2007年第4期16-19,共4页
Applied Science and Technology
关键词
语音编码
线性预测
脉冲散布
码本搜索
speech coding
linear prediction
pulse-dispersion
codebook research