摘要：AbstractIn order to detect the difference of acoustic features in effective on emotional perception such as happiness, sadness, anger and fear between Japanese and Chinese. In addition, we also adopted some impressions as the middle layer of these estimation models by two steps. First, we produced a linear estimation model for each emotions with the values of impressions evaluated, and we also created the models to estimate impressions. We assessed the models created in different way on its accuracy and the standardized partial regression coefficients. As a result, happiness and anger show that they have same tendency on acoustic features contributed in estimating impressions. We assumed that they have little or none cultural dependency on perception of impressions at least between Japanese and Chinese.
