Woongjib Choi

Hello! I am a researcher at the Digital Signal Processing and AI (DSPAI) Lab at Yonsei University (South Korea). I received my B.S. in Electrical and Electronic Engineering from Yonsei University in 2024, and completed my M.S. at the same university in February 2026 under Prof. Hong-Goo Kang. My primary research goal is to bridge the gap between classic signal processing and modern deep learning, with interests spanning digital signal processing, speech and audio processing, audio compression, and generative models.

My first research experience as an undergraduate at the SRIP Lab involved implementing and analyzing traditional image signal processing algorithms such as denoising, demosaicing, and super-resolution. This interest in combining established algorithms with data-driven approaches has continued throughout my graduate studies.

During my M.S., I focused on neural audio codecs and audio super-resolution. Specifically, I worked on non-blind bandwidth extension using side information for legacy audio codec enhancement, as well as blind audio super-resolution using flow matching, where both approaches are grounded in time-frequency and subband analysis. Currently, my research interests extend to audio tokenization and generative models for audio, with additional experience in multilingual speech processing.

News

Mar 04, 2026	📝 Submitted two papers to INTERSPEECH 2026: “SAGE-LD” and “UR-BERT”. Fingers crossed! 🤞
Feb 23, 2026	🎓 Successfully completed my M.S. in Electrical and Electronic Engineering at Yonsei University!
Jan 16, 2026	🎉 My first-author work, “UniverSR: Unified and Versatile Audio Super-Resolution via Vocoder-Free Flow Matching”, has been accepted to ICASSP 2026. Excited to present this work in Barcelona! 😊
Sep 22, 2025	🎼 Attending ISMIR 2025 in Daejeon was fascinating opportunity to connect with researchers in the music and audio domains!
Aug 18, 2025	🎤 I presented my first-author paper “Neural Spectral Band Generation for Audio Coding”, at INTERSPEECH 2025 in the Netherlands. What a great experience! 😊

Selected Publications

ICASSP

UniverSR: Unified and Versatile Audio Super-Resolution via Vocoder-Free Flow Matching

Woongjib Choi, Sangmin Lee, Hyungseob Lim, and 1 more author

In IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2026

Abs Code LINK Demo

In this paper, we present a vocoder-free framework for audio super-resolution that employs a flow matching generative model to capture the conditional distribution of complex-valued spectral coefficients. Unlike conventional two-stage diffusion-based approaches that pre dict a mel-spectrogram and then rely on a pre-trained neural vocoder to synthesize waveforms, our method directly reconstructs waveforms via the inverse Short-Time Fourier Transform (iSTFT), thereby eliminating the dependence on a separate vocoder. This design not only simplifies end-to-end optimization but also overcomes a critical bottleneck of two-stage pipelines, where the final audio quality is fundamentally constrained by vocoder performance. Experiments show that our model consistently produces high-fidelity 48 kHz audio across diverse upsampling factors, achieving state-of-the-art performance on both speech and general audio datasets.
INTERSPEECH

Neural Spectral Band Generation for Audio Coding

Woongjib Choi, Byeong Hyeon Kim, Hyungseob Lim, and 2 more authors

In Proceedings of INTERSPEECH, 2025

Abs LINK

Spectral band replication (SBR) enables bit-efficient coding by generating high-frequency bands from the low-frequency ones. However, it only utilizes coarse spectral features upon a subband-wise signal replication, limiting adaptability to diverse acoustic signals. In this paper, we explore the efficacy of a deep neural network (DNN)-based generative approach for coding the high-frequency bands, which we call neural spectral band generation (n-SBG). Specifically, we propose a DNN-based encoder-decoder structure to extract and quantize the side information related to the high-frequency components and generate the components given both the side information and the decoded core-band signals. The whole coding pipeline is optimized with generative adversarial criteria to enable the generation of perceptually plausible sound. From experiments using AAC as the core codec, we show that the proposed method achieves a better perceptual quality than HE-AAC-v1 with much less side information.