Hanyang University (Mar., 2011 - Aug., 2016)
Seoul National University (Sep., 2016 - Feb., 2022)
Thesis Award
Title: “A Controllable Generation of Signals from Self-Supervised Representations”
2024 Mar. -
2020 Mar. - 2024 Feb.
2018 Summer
(May 2023, ICLR)
NANSY++: Unified Voice Synthesis with Neural Analysis and Synthesis
(December 2021, NeurIPS)
Neural Analysis and Synthesis: Reconstructing Speech from Self-Supervised Representations
(June 2021, ICASSP)
Real-time Denoising and Dereverberation with Tiny Recurrent U-Net
(May 2020, Arxiv)
Phase-aware Single-stage Denoising and Dereverberation with U-Net
(April 2020, ICLR)
From Inference to Generation: End-to-end Fully Self-supervised Generation of Human Face from Speech
(November 2019, ISMIR)
Audio-query Based Source Separation
(April 2019, Arxiv)
Phase-aware Speech Enhancement with Deep Complex U-Net
(December 2017, Neurips ML4Audio Workshop (Oral presentation))
Singing Voice Separation using Generative Adversarial Networks
(2022, Transactions on Audio, Speech, and Language Processing) **
Differentiable Artificial Reverberation
(November 2020, ISMIR)
Exploring Aligned Lyrics-informed Singing Voice Separation
(May 2020, ICASSP)
Disentangling Timbre and Singing Style with Multi-Singer Singing Synthesis System
(September 2019, Interspeech (Best paper award))
Adversarially Trained End-to-end Korean Singing Voice Synthesis System
(October 2023, WASPAA)
AECSQI: Referenceless Acoustic Echo Cancellation Measures using Speech Quality and Intelligibility Improvement
Yet Another Generative Model for Room Impulse Response Estimation
(June 2023, ICASSP)
Towards Trustworthy Phoneme Boundary Detection With Autoregressive Model and Improved Evaluation Metric
(2023, IEEE International Solid-State Circuits Conference (ISSCC))
A 0.81mm2 740µW Real-Time Speech Enhancement Processor Using Multiplier-Less PE Arrays for Hearing Aids in 28nm CMOS
(June 2021, ICASSP)
Room Adaptive Conditioning Method for Sound Event Classification in Reverberant Environments
CES2022
I developed a controllable voice conversion technology that can convert arbitrary speech and singing voice into any target voice in real-time.
ICASSP2021, Microsoft
I developed a speech enhancement algorithm as a lead of the team SNU/Supertone. I won 3rd prize in the ICASSP2021 Deep Noise Suppression (DNS) challenge out of 16 participants from diverse universities and companies using about 10 times fewer parameters than the first and second winners.
Interspeech 2019
We won the best student paper at Interspeech 2019 by proposing a neural singing voice synthesis system.
Samsung Advanced Institute for Health Sciences & Technology, Nov 2018
We won the first prize in the hackathon by developing a deep learning-based hearing-aid framework and by suggesting a future concept of digital hearing-aid.
Conference Session Chair
Interspeech: 2022
Reviewer
NeurIPS: 2020, 2021 (outstanding reviewer), 2022, 2023
ICML: 2021, 2022
ICLR: 2020, 2021, 2022
ISMIR: 2021, 2022
Interspeech: 2022, 2023
JSTSP: 2018
IEEE Signal Processing Letter: 2021
Volunteer
International Conference Tutorial: Designing Controllable Synthesis System for Musical Signals @ ISMIR2022
T3(M): Designing Controllable Synthesis System for Musical Signals
Controllalble Synthesis of Speech and Singing Voice with Neural Networks @ Korea Advanced Institute of Science and Technology (KAIST, GCT634, Musical Applications of Machine Learning), Nov. 29, 2021
Deep Learning Based Speech Enhancement @ Korea Institute for Advanced Study (KIAS), Dec. 12, 2019
Conditional Generative Model for Audio @ Soundly (Acoustic Speech Audio Language Processing Community in Korea), Nov. 30, 2019