Gyubin Lee

MS student @ KAIST MAC Lab (Advisor: Juhan Nam)

Research Interest: General Audio Generation, Multi-modal LLM, Audio Signal Processing, Reinforcement Learning

✉️ gbstorm81@kaist.ac.kr
🎓 Google Scholar
🏢 LinkedIn

Education

Incheon Academy of Science & Arts (IASA)
2016 Mar - 2019 Feb

Korea Advanced Institute of Science and Technology (KAIST)
BS: School of Computing, GPA: 3.92/4.30
2019 Feb - 2025 Aug

Technical University of Denmark (DTU)
Exchange student
2024 Aug - 2024 Dec

Korea Advanced Institute of Science and Technology (KAIST)
MS: Graduate School of AI
2025 Sep -

Publications

CONMOD: Controllable Neural Frame-based Modulation Effects
Gyubin Lee, Hounsu Kim, Junwon Lee and Juhan Nam
Proceedings of the International Conference on Digital Audio Effects (DAFx), 2024

*** Best Paper Awarded ***
[paper][demo][news]

Memory-guided goal-driven reinforcement learning explains subclinical depression
Gyubin Lee, Minsu Abel Yang and Sangwan Lee
Proceedings of the Multi-disciplinary Conference on Reinforcement Learning and Decision Making (RLDM), 2022

Research Experience

Music and Audio Computing Lab
Research Intern (Individual Study)
Advisor: Prof. Juhan Nam
Korea Advanced Institute of Science and Technology | 2023 Sep - 2024 Jul

Brain x Machine Intelligence Lab
Research Intern (Undergraduate Research Program, Individual Study)
Advisor: Prof. Sangwan Lee
Korea Advanced Institute of Science and Technology | 2020 Dec - 2022 Feb

Certificate & Awards

KAIST Presidential Fellowship (KPF)
2019 Feb

DAFx 2024 Best Paper
2024 Sep

Projects

FairSpeech

Speech Emotion Recognition (SER) models are often biased due to gender-based expression differences and imbalanced datasets. To address this, we propose FairSpeech, a Transformer-based model that uses adversarial bias mitigation to significantly reduce gender bias while maintaining competitive performance. FairSpeech works by explicitly minimizing the emotion loss disparity between male and female speakers.

Musical reCAPTCHA

This website features a unique musical reCAPTCHA challenge where users must reproduce a sequence of notes. To begin, you click "Play Challenge" to listen to a 5-note melody. You then play the notes back in the correct order using the provided musical keyboard to prove you are a human.

Page updated

Google Sites

Report abuse