改进参数控制的可视语音合成方法

Improved visual speech synthesis method of parameter control

传统单音素对音节内部和音节之间的协同发音影响采用相同处理方法,为此,分析音节内部和音节之间两种不同协同发音现象对可视语音合成的不同影响,提出一种改进参数控制的可视语音合成方法。针对不同音节,不改变元、辅音视位峰值处权值函数的幅度及其变化速度参数,仅修改元、辅音的时间参数,使修改后的元、辅音视位参数能更好地模拟真实音节发音过程中发音器官的动态变化特征。实验结果表明,改进方法能有效地解决音节内协同发音的问题,改善了可视语音合成的质量。

To solve the unreasonable problem that the same processing method was adopted for coarticulation problem with inner syllabic and inter-syllabic, the different influences on the visual speech synthesis of different kinds of coarticulation phenomenons with inner-syllabic and inter-syllabic were analyzed, and an improved visual speech synthesis method based on monophone para- meter control was put forward. The amplitude of the weight function about the peak value and the rate of change of the vowels viseme and consonants viseme were unchanged, only the time parameters of vowels viseme and consonants viseme were modified. The modified parameters were used to simulate the dynamic change characteristics in the process of real syllable pronunciation organs, bringing about better performances. The feasibility of the method was validated by practical application. Results show that the improved method can solve the coarticulation problem with inner-syllabic, and improve the quality of the visual speech synthesis.