对自然语音和人工语音所诱发的情绪感知差异的ERP研究

An ERP Study of Emotional Perception Differentiated by Natural and Artificial Voice

采用事件相关电位对人脑在相同情绪效价的人工语音和自然语音(中立)下情绪处理进行探索,以及研究人工语音与其他情绪效价的自然语音(欢乐、生气)在脑电信号上有何区别。对29名大学生采用过德国Brain Products公司的脑电记录系统采集,按照10-20系统扩展的64导电极帽记录在带有情绪的语音启动下的ERP,脑电系统记录其在情绪语音的条件下的事件相关电位和行为学结果。在700~800ms时间段内,人工语音比自然语音(欢乐、中立、生气)引发了更正走向的波幅(P<0.05);大学生对相同效价的人工语音和自然语音确实存在一定的感知差异,在700-800ms的时间窗口里,人脑正在对人工语音进行处理,对于人工语音的资源分配显著多与其他语音的资源分配。

This study aimed to identify the differences in emotional processing between artificial voices and natural human voices(neutral)by using event related potential(ERP)and to clarify how different between artificial voices and natural human voices(happy、mad)in ERP. The ERP elicited under different emotional voices conditions were recorded of 29 university students and their behavioral performance was collected. The event-related EEG signals were recorded and analyzed to find out the appropriate time window of ERP. During the 700-800 ms time window, artificial voices aroused more large positive amplitudes(LPP or LPC)than natural human voices(P<0.05). Participants were most sensitive to neutral intonation of natural human voices. Artificial voices lead to more positive LPP or LPC, the emotional processing of artificial voices was slower than neutral intonation of natural human voices. The resource allocation of artificial voices was much more than other kinds of natural human voices when brains addressed voices. The speed of addressing neutral intonation of natural human voices for the brains was the fastest and it consumed the least cognitive resources.