基于基音同步帧叠接的吴语语音合成

Synthesis of Wu-Dialect Speech Based on Pitch- Synchronous Frame Concatenation

吴语语音和普通话一样,也是由声母、韵母和声调这三个要素构成。语音的平稳段存在于韵母中,波形具有周期性,周期的变化趋势决定语音的声调;非平稳段包括声母、过渡音和鼻音。根据这些特点,提出了一种利用基音同步帧和声调曲线来合成语音的新方法PSFC,并用该方法合成吴方言。实验表明,该方法简单实用,合成语音有较好的自然度和可懂度。

As mandarin, the Wu-dialect speech consists of three factors that are initial, final and tone. In Chinese speech, the stable section was included in the final part . The waveform of the stable section appears evidently periodic and the tone of the speech is decided by the change of period. The unstable section consists of initial, transitional sound and nasals . This paper proposed a new method of speech synthesis that synthesizes speech using pitch-synchronous frame and tone contour. According to the experiments of Wu-dialect speech synthesis, the method is quite concise and practicable, the synthesisresult is satisfied in the aspects of naturalness and intelligibility.