摘要
提出了一种基于最佳相位设计的语音合成技术,能够有效降低MBE声码器合成语音信号由于波形失衡而导致的饱和失真的概率。此外,为了保证合成滤波器的稳定性,对线谱频率(LSF)系数提取进行了优化。实验结果显示,合成语音信号波形近似平衡地分布在零幅度值的上下,语音听起来没有不舒服的感觉。实验结果表明,基于最佳相位设计的语音合成技术能够有效改善合成语音质量。
A speech synthesis technology based on optimal phase design is proposed. It can effectively reduce the chance of saturation distortions caused by the unbalanced waveform for the synthesized speech signal in the MBE vocoder. Moreover, to ensure the stability of the synthesizing filter, the line spectral frequency （LSF） coefficient extracting is optimized. The experimental results show that the synthesized speech signal waveform displays an approximate balanced distribution up and down from the zero amplitude value, and that the speech sounds without uncomfortable feeling. It is justified from the experimental results that the synthesized speech quality can be effectively improved by using the proposed speech synthesis technology based on the optimal phase design and the optimized LSF coefficient extracting.
出处
《计算机与数字工程》
2012年第9期21-23,35,共4页
Computer & Digital Engineering
关键词
语音合成
多带激励
最佳相位
线谱频率
speech synthesis
multiband excitation
optimal phase
line spectral frequency
