论文部分内容阅读
本文选择宋词为实验语料,使用PowerLab生物电采集器、MLT1132压电呼吸带传感器等仪器录制呼吸信号和语音信号,对语音信号的韵律等级进行切分,通过判断呼吸周期的波峰和波谷来标记吸气相和呼气相,揭示了不同语音韵律等级时长和两个常见的呼吸结构模型之间的相关性。在语音多模态研究中,可以为语音多模态的合成提供可靠的数据,提高言语模型的自然度。准确的数据模本还可以为治疗言语障碍提供帮助,为语言教学和二语习得提供标准。
In this paper, Song words are selected as the experimental corpus, respiratory signals and speech signals are recorded by instruments such as PowerLab bioelectric collector and MLT1132 piezoelectric breathing zone sensor, and the prosody level of the speech signal is segmented, and the peak and trough of the respiratory cycle are judged to be labeled Gas phase and expiratory phase reveal the correlation between different prosody levels and two common respiratory structural models. In the multi-modal study of speech, it can provide reliable data for multi-modal synthesis and improve the naturalness of speech model. Accurate data models can also help to treat speech impairments, providing standards for language teaching and second language acquisition.