基于提升小波分解的低速率波形内插语音编码算法

提出了一种基于双正交提升小波变换(bi-orthogonal lifting wavelet transform,BLWT)的低速率特征波形内插语音编码方法,其中的特征波形分解算法不需要复杂的特征波形对齐操作和滤波器的卷积运算,其固有的原位运算降低了传统特征波形小波分解算法所需的内存,当前帧边界点替代相邻帧样点的措施有效减少了传统特征波形小波分解算法的时延.同时,该分解方法对分解后的各成分单独重建,并根据人耳的感知特性选择量化参数.基于该分解,分别构建了1.84 kb/s和2.32 kb/s两种速率的BLWT-CWI(characteristic waveform interpo-lation)语音编码器.主观平均意见得分(mean opinin score,MOS)结果表明,2.32 kb/s的BLWT-CWI语音编码质量与2.4 kb/s的MELP声码器相当,1.84 kb/s的BLWT-CWI语音编码质量稍逊于2.4 kb/s的MELP声码器.主观A/B听力测试结果表明,1.84 kb/s的BLWT-CWI语音编码质量优于2 kb/s的LIWI(low-complex improvedwaveform interpolation)声码器.

Waveform Interpolation Speech Coding Algorithm at Low Bit Rates on the Basis of Lifting Wavelet Decomposition

A characteristic waveform interpolation(CWI) speech coding algorithm at low bit rates based on bi-orthogonal lifting wavelet(BLWT) is proposed in this paper.The complicated characteristic waveform(CW) alignment operation and convolution operation of filter are cancelled by using BLWT-based CW decomposition.The memory of traditional CW wavelet decomposition algorithm is reduced with its inherent situ calculation.The algorithm delay of the traditional CW decomposition based on wavelet transform is decreased by replacing the samples of adjacent frames with the boundary samples of the current frame.Each decomposed component of CW is independently reconstructed and the quantization parameters of CW are flexibly selected according to the human ear’s perception.Two types of CWI speech codecs at 1.84 kb/s and 2.32 kb/s are designed based on BLWT.MOS test results show that 2.32 kb/s BLWT-CWI and 2.4 kb/s MELP have similor quality and the performance of 1.84 kb/s BLWT-CWI is slightly less than 2.4 kb/s MELP.Subjective A/B listening tests also show that the quality of 1.84 kb/s BLWT-CWI is better than that of 2 kb/s LIWI(low complex improved waveform interpolation)codec.