论文部分内容阅读
声码器因其极低的语音压缩编码速率而得到广泛应用。但是声码器自身的内在算法特性使其通常对系统的运算量和存储量要求很高,这就为其在某些场合的应用造成了困难。为了解决以上问题,提出了一种基于正弦激励线性预测(SELP)模型的2.4 kb/s低复杂度高质量声码器算法。新提出的声码器在原有SELP声码器算法基础上采取了一系列有针对性的改进措施,在大幅缩减算法时间复杂度和空间复杂度的同时,保证了很高的合成语音质量。
Vocoder is widely used because of its very low voice compression coding rate. However, the inherent algorithm characteristics of the vocoder usually require high computational complexity and storage capacity of the system, which makes it difficult to use in some occasions. In order to solve the above problem, this paper proposes a 2.4 kb / s low complexity high quality vocoder algorithm based on sinusoidal excitation linear prediction (SELP) model. The new proposed vocoder based on the original SELP vocoder algorithm to take a series of targeted improvements to significantly reduce the algorithm time complexity and space complexity at the same time, to ensure a high synthetic speech quality.