Search

Due to the influence of culture and society on the expression of human emotion, the features of speech emotion in different languages vary greatly, which leads to the insufficient generalization ability of speech emotion recognition model in a single language . To solve this problem, this paper proposed a multi-lingual speech emotion recognition method based on multitask attention. By introducing the auxiliary task of language identification, the model could not only learn the emotional features shared by different languages, but also learn the unique emotional characteristics of each language, so as to improve the generalization ability of the multi-language emotion recognition model. Experiments on the dimensional affective corpora of two languages show that the proposed method improves the mean values of relative UAR of the Valence and Arousal tasks by 3 . 66% ~ 5. 58% and 1. 27% ~ 6. 51 %, respectively, compared with the benchmark methods . Experiments on discrete affective corpora of four languages show that the mean values of absolute UAR improves by 13. 82% ~ 15. 75% compared with the benchmark methods. Therefore, the proposed method can effectively extract the language-related emotion feat ures and improves the performance of multi-lingual emotion recognition. [ABSTRACT FROM AUTHOR]