[1H5-GS-10] Vision, speech: Recognition and detection

[1H5-GS-10-03] Research on Relevance of Different Languages Dealing with Emotion Recognition on the Basement of Audio Signal Processing

DING AN¹, 〇INOUE SATORU¹ (1. Saitama Institute of Technology)

Keywords:AI, audio signal processing, emotion recognition

Speaking of audio recognition, the words spoken by the speaker are often converted into visible characters. However, it is difficult to figure out the speaker’s true intention only looking at letters, and the conversation may not proceed smoothly. In order to understand the true intention of the speaker more accurately, it is necessary to identify one’s emotions in voices. This study used openSMILE to extract various emotional features from the voice data of emoDB, and tried to classify them by SVM. In addition to verifying the prediction accuracy in the same language, this study also examined whether emotional expressions were related between different languages. The purpose to evaluate emotional components including voice will lead to machine learning in the future.