【作者】 陆灵犀； 包小涵； 吴玺宏； 李量；

【Author】 Lingxi Lu;Xiaohan Bao;Xihong Wu;Liang Li;Department of Psychology,Speech and Hearing Research Center,Key Laboratory on Machine Perception Ministry of Education,Peking University;

【机构】 北京大学心理学系智能科学系言语和听觉研究中心,机器感知与智能重点实验室；

【摘要】 在嘈杂环境中,大脑能够整合目标嗓音的多种物理学和心理学特征,促进对目标言语的加工和理解。本研究由两个实验组成,探讨嘈杂环境下目标嗓音的情绪学习和主观空间分离的去掩蔽作用。实验一采用情绪学习分化厌恶条件化范式,在两个说话人言语掩蔽的环境中测量4个信噪比条件下对目标嗓音(负性情绪学习条件CS+和中性学习条件CS-)的言语识别正确率,并通过心理物理曲线拟合得到言语可懂度阈限。结果发现,通过负性情绪学习后的目标嗓音识别正确率显著高于中性学习后的嗓音识别正确率,负性嗓音可懂度阈限显著低于中性嗓音。情绪和主观空间分离两因素无交互作用,两者共同的去掩蔽作用接近于两者单独作用的线性叠加。实验二中掩蔽刺激为稳态语谱噪声,结果发现情绪的去掩蔽作用消失。这说明情绪学习能够作用于去信息掩蔽过程,而不作用于去能量掩蔽过程。通过统计对负性和中性嗓音的情绪主观评定值,发现无论言语掩蔽还是噪音掩蔽,被试对嗓音的负性情绪信息越敏感,情绪学习对言语识别的促进作用越强。更多还原

【Abstract】 In noisy environment,our brain integrates varieties of physical and psychological characteristics of the target voice to promote the processing and understanding of the target speech.The present study consists of two experiments which explored the unmasking effect of emotional learning and subjective spatial separation of the target voice in noisy environment.In experiment one,we adopted the aversive learning paradigm to test the recognition of the target voices(negative emotional learning condition CS+ and neutral learning condition CS-) in four SNRs under speech masking condition,and fitting the recognition accuracy using psychophysical function to obtain the intelligibility thresholds in different conditions.The results indicated that the accuracy of speech recognition was higher for the negative learning targets compared to the neutral ones,and the intelligibility threshold of negative voices was lower than neutral voices.No interaction was found betweenthe two factors of emotion and subjective spatial separation.The combined unmasking effect of the two factors was nearly the linear sum of their single effects.In experiment two,the masking stimulus was steady-state speech-spectral noise.The results showed no significant unmasking effect of emotion,which indicated that emotional learning interacted with informational masking more than energetic masking.The analysis of Self-Assessment Manikin ratings on emotional and neutral voices suggested that the more sensitive participants showed to the negative information of the voices,the greater emotional learning enhanced speech recognition.更多还原