Sattaya Singkul received the B.Sc. in information technology from King Mongkut's Institute of Technology Ladkrabang (KMITL) in 2019. During 2019-2021, he received a scholarship to pursue a master's degree in innovator's promising with honor and currently pursuing the M.Sc. in information technology from KMITL. Now he is a researcher in speech and text understanding laboratory at Thailand's National Electronics and Computer Technology Center (NECTEC). Besides, he is co-researcher within cross-functional teams such as Siam Commercial Banking (SCB), Kasikorn Business Technology Group (KBTG), Thailand Mental Health Technology and Innovation Center in Mahidol University, KMITL, and Artificial Intelligence Association of Thailand (AiAT) to expanding in Thailand and globally. His research interests are in speech processing, music processing, language understanding, intelligent systems, and healthcare systems.
Nowadays, emotion analysis has been an active research area in human-computer interactions which depend on language understanding. For Thai, language understanding has been many challenges from cultural language that is shortened words, ambiguous words, slangs, sarcastic meaning, and homophones. Besides, Thai is a low-resource language that has low dataset size and research when compare with high-resource language. For Thai emotion analysis, Thai speech emotion recognition (SER) is same challenges as previously mentioned. Additionally, Thai speech emotions are individually speech person style and have a variety of contexts. All of which cause Thai to be difficult to handle especially in SER tasks. Therefore, in this study, we would like to present the Thai SER challenge with solving solutions using deep learning approach. The study outlines are described in three sections including language analysis, SER feature analysis, and SER model to improved Thai SER performance.