基于奇异值分解的低速率波形内插语音编码算法

波形内插(WI)语音编码模型作为当今最具潜力的低速率语音编码方案之一,因其良好的性能,越来越受到人们的重视.本文基于一种奇异值分解(SVD)的特征波形分解方法,利用语音信号的感知特性,将二维特征波形的幅度谱分成基本矩阵、过渡矩阵和补充矩阵,并采用了不同的量化方法,有效地降低了运算复杂度;另外,本文根据语音信号时变特性,将三个矩阵分为三种组合模式表示特征波形幅度谱,并引入周期因子和能量熵来衡量矩阵周期程度,解决了奇异值分解后参数难于量化的问题,提高了编码效率.主观A/B测试表明,本文提出的2.4kbps SVD-WI编码器的重建语音质量略好于2.4kbps MELP编码器.

Low Bit Rates Waveform Interpolation Speech Coding Based on Singular Value Decomposition

As one of the current most potential low bit rates speech coding schemes, waveform interpolation model with its high performance,has been paid more and more attention. Based on a kind of Singular Value Decomposition method for decomposing Characteristic Waveform in WI speech coding, the magnitude spectrum of CW is perceptually divided into basic matrix, transitional matrix and supplemental matrix to be quantized respectively, which effectively reduce the computational complexity; In addition, there are three patterns to represent the magnitude spectrum of CW combined by the matrixes above, and then Periodic Factor and Energy Entropy are introduced to indicate the periodicity of matrix, which solved the problems for quantization of parameters after SVD and improved the efficiency of coding. Subjective A/B listening tests indicated that the reconstructed speech quality of the 2.4kbps SVD-WI codec is a little better than that of 2.4kbps MELP coder.