In the preprocessing stage, the audio data is recognized into the syllable graph with high accuracy syllable candidates.

• 预处理阶段将语音数据识别成具有高覆盖率的拼音图。