[LAB Search] Leading Voice and Sound AI Technology... Research Team of Professor Jang Junhyuk at Hanyang University

Professor Junhyuk Jang's research team from the Department of Convergence Electronics Engineering at Hanyang University won first place in the audio generation category of the global acoustic artificial intelligence (AI) competition, the ‘IEEE DCASE 2023 Challenge’.

Professor Jang Jun-hyuk of Hanyang University (far left) and members of the Acoustic Signal Processing and Machine Learning Laboratory.
[Photo by Hanyang University]

Professor Jang's team took first place in the 'Audio Generation' category, where 14 teams worldwide submitted 28 systems, and second place in the 'Automatic Audio Captioning' category, which featured 10 teams presenting 29 systems. Notably, the technology used in the audio generation field was designed by utilizing diffusion-based technology, which is emerging in generative AI, and the representative generative model 'GAN' (Generative Adversarial Networks). By combining the strengths of these two technologies, the team received high evaluations for generating not only high-quality audio but also a variety of sounds.

Professor Jang's team is dedicated to advancing the speech and acoustic fields through active research, including experiments on various datasets and the introduction of the latest deep learning algorithms. The Speech and Acoustic Signal Processing and Machine Learning Laboratory, led by Professor Jang, conducts diverse research on deep learning-based speech, acoustics, and signal processing. It is a large-scale laboratory rare even globally, consisting of 25 doctoral students, 17 master's students, and 5 interns. They form teams according to research areas such as speech recognition, speech synthesis, speaker recognition, and signal processing. The lab has proven its capabilities by publishing numerous papers in internationally top-tier conferences recognized in the speech audio AI field, such as 'ICASSP' and 'INTERSPEECH.' Additionally, they maintain steady communication with leading domestic companies like Samsung Electronics and Hyundai Motor Company, conducting industry-academic projects and seminars.

Currently, AI technology development research in the speech and acoustic fields is actively progressing worldwide because communication between AI and humans is essential to replace human tasks. This is why global companies such as OpenAI, Google, and Microsoft are introducing and investing in data learning models like ChatGPT. Professor Jang said, "Our lab members are working together diligently on paper submissions for ‘ICASSP 2024,’ an internationally renowned conference in the speech AI field to be held in Seoul next year," adding, "We plan to continuously strive to quickly acquire the latest technology trends and nurture researchers who are competitive not only domestically but also globally."