论文部分内容阅读
为了解决码本激励线性预测(CELP)算法在4kb/s以下质量难以提高的问题,在编码器和解码器中分别引入了不同的脉冲散布技术以达到利用较少比特提供高质量激励的目的。结合固定码本基音增强和脉冲散布技术的特点,根据理想代数码本脉冲在子帧中各位置的概率分布规律,提出了新的非均匀结构代数码本,明显地提高了代数码本的效率。同时引入了基音预加重技术以提高自适应码本的搜索精度。经主观听音和计算机模拟测试,表明最后形成的3.35kb/s多重脉冲散布非均匀代数码本激励线性预测(MPD-USACELP)语音编码算法的质量达到了北美IS-54标准8kb/s算法的水平。
In order to solve the problem that the CELP algorithm can not improve the quality under 4kb / s, different pulse spreading techniques are introduced in the encoder and decoder respectively to achieve the purpose of providing high quality excitation with fewer bits. According to the characteristics of pitch enhancement and pulse spreading techniques of fixed codebook, a new non-uniform structure algebraic codebook is proposed according to the probability distribution rule of ideal algebraic code pulse in every position of sub-frame, which obviously improves the efficiency of algebraic codebook . At the same time, Pitch pre-emphasis technique is introduced to improve the searching accuracy of adaptive codebook. Subjective listening and computer simulation tests show that the quality of the finally formed 3.35kb / s multi-pulse-spread non-uniform algebraic codebook motivational linear prediction (MPD-USACELP) speech coding algorithm achieves the quality of the North American IS-54 standard 8kb / s algorithm Level.