GIST AiTeR

People

Members

Meet the PI and current researchers.

Prof. Hong Kook Kim
Professor, Department of Electrical Engineering and Computer Science · Adjunct Professor, Department of AI Convergence
EECS Building C, Room 507
hongkook@gist.ac.kr

Curriculum Vitae

김홍국 Hong Kook Kim

Professor

Affiliation & Contact

Affiliation
Department of Electrical Engineering and Computer Science,
Department of AI Convergence (Adjunct),
Gwangju Institute of Science and Technology (GIST)
E-mail
hongkook@gist.ac.kr
Tel
062-715-2228
Fax
062-715-2204
Office
전기전자컴퓨터공학부 C동 507호

Education

  • 1994Ph.D., Electrical Engineering — KAIST
  • 1990M.S., Electrical Engineering — Korea Advanced Institute of Science and Technology
  • 1988B.S., Control and Instrumentation Engineering — Seoul National University

Work Experience

  • 2017Dean of Planning — GIST
  • 2015 – 2017Dean of Electrical Engineering and Computer Science — GIST
  • 2008 – presentProfessor — GIST
  • 2003 – 2008Associate Professor — GIST
  • 1998 – 2003Senior Technical Staff Member — AT&T Labs-Research, Florham Park, New Jersey
  • 1998Senior Engineer — MMC Technology, Inc., Seoul, Korea
  • 1990 – 1998Senior Researcher — Samsung Advanced Institute of Technology, Kiheung, Korea

Research Fields

Speech Enhancement & Audio Coding

  • Noise reduction for improvement of speech quality and recognition performance
  • De-reverberation
  • Acoustic echo cancellation and residual echo suppression
  • Audio and speech coding

Speech and Sound Event Recognition

  • Speech recognition based on statistical and deep learning models
  • Pronunciation and language modeling
  • Acoustic event and scene detection based on deep learning classification models
  • Speech synthesis from input text by end-to-end synthesis systems

3D Audio

  • Stereo audio channel to 5.1 channel upmixing
  • Head-related transfer function modeling for personalized audio experience
  • Sound source separation across various objects
  • Sound source localization

Climate Change Prediction

  • Climate factor prediction based on recurrent neural networks
 Scholar LinkedIn

Work Experience

  • 2017Dean of Planning — GIST
  • 2015 – 2017Dean of Electrical Engineering and Computer Science — GIST
  • 2008 – presentProfessor — GIST
  • 2003 – 2008Associate Professor — GIST
  • 1998 – 2003Senior Technical Staff Member — AT&T Labs-Research, Florham Park, New Jersey
  • 1998Senior Engineer — MMC Technology, Inc., Seoul, Korea
  • 1990 – 1998Senior Researcher — Samsung Advanced Institute of Technology, Kiheung, Korea
AunionAICompany

Current Researchers

CurrentAlumni

Ph.D. Student

Jeonghyeok Lee
Department of Electrical Engineering and Computer Science
ljh0412@gist.ac.kr

Curriculum Vitae

이정혁 Jeonghyeok Lee

Ph.D. Student

Affiliation & Contact

Affiliation
Gwangju Institute of Science and Technology (GIST), Gwangju, Republic of Korea

Research Interests

My research focuses on developing and optimizing deep learning models for speech recognition, text-to-speech synthesis, and audio signal analysis, with specific applications in healthcare diagnostics, call center efficiency, and secure communications.

Publications

Domestic

  • 이정혁, 이건우, 김성주, 봉귀영, 유희정, 김홍국 “깊은 신경망을 활용한 음성 기반의 자폐스펙트럼 장애 판별에 관한 기초 연구,.” 2020년도 한국통신학회 하계종합발표회논문집, pp. 66 (2020)
  • 이정혁, 홍찬기, 김홍국 “저용량 파라미터 튜닝에 기반한 Text-to-Speech 모델 최적화,.” 2023년 한국통신학회 하계종합학술발표회 논문집, pp. 1600-1601 (2023)
  • 정승훈, 권오현, 나형주, 마승희, 전슬기, 오현식, 이정혁, 김나연, 김홍국 “콜센터 데이터셋에서 음성인식 모델의 연속학습 연구,.” 2023년 한국통신학회 하계종합학술발표회 논문집, pp. 1594-1595 (2023)
  • 정승훈, 권오현, 나형주, 마승희, 전슬기, 오현식, 이정혁, 김나연, 김홍국 “한국어 음성인식을 위한 시퀀스 단위 연속학습 및 콜센터 적용,.” 2023 한국음성학회 봄 학술대회 논문집 (2023)
  • 마승희, 권오현, 나형주, 정승훈, 이정혁, 박현주, 김홍국 “긴 발화 콜센터 데이터에 적합한 한국어 음성 인식 모델 연구,.” 2022년 한국통신학회 하계종합학술발표회 논문집, pp. 1821-1822 (2022)
  • 이정혁, 김홍국 “1차원 합성곱 신경망 구글넷 기반 스테레오 음성의 도래 방향 추정,.” 2021년 한국음성학회 가을 학술대회 논문집, pp. 25 (2021)
  • 김나연, 이정혁, 박창수, 김홍국 “광섬유 OTDR 데이터의 이상 상태 감지를 위한 오토인코더 구조 비교 연구,.” 2021년 한국통신학회 추계종합학술발표회 논문집 (2021)
  • 이정혁, 김선교, 김홍국 “보안 통신용 보코더에 적용 가능한 자기부호화기 기반 스펙트럼 변형,.” 2020년도 한국전자파학회 하계종합학술대회 논문집, pp. 268 (2020)
  • 이정혁, 김홍국, 김선교 “딥 러닝 오토인코더 기반 보코더 파라미터 감축,.” 2019 한국음성학회 봄 학술대회 발표 논문집, pp. 131 (2019)
  • 이정혁, 김홍국 “보코더 비트스트림을 이용한 딥러닝 기반 군용 보코더 유형 검출,.” 한국음성학회 2018년 봄 학술대회 논문집, pp. 87 (2018)
  • 이건우, 이정혁, 양정현, 김홍국, 안충현 “메인 및 서브 오디오의 켑스트럼 파라메터를 이용한 화면해설 오디오 구간 검출,.” 한국음성학회 2018 봄 학술대회 논문집, pp. 26 (2018)
Jimin Jeon
Department of Electrical Engineering and Computer Science
jiminbot20@gm.gist.ac.kr

Curriculum Vitae

전지민 Jimin Jeon

Ph.D. Student

Affiliation & Contact

Affiliation
Gwangju Institute of Science and Technology (GIST), Gwangju, Republic of Korea

Research Interests

My research interests focus on developing deep learning models for speech emotion recognition and short-term temperature forecasting using advanced embedding and hybrid neural network architectures.

Publications

Domestic

  • 전지민, 김홍국 “Wav2Vec 2.0 기반 임베딩을 활용한 한국어 음성감정인식,.” 제32회 신호처리합동학술대회 논문집, pp. 254-256 (2022)
  • 전지민, 김홍국 “LDAPS 예측 및 AWS 관측 데이터를 이용한 CNN-BLSTM 기반의 단기 기온 예측,.” 2021년 제2회 한국인공지능학술대회 논문집, pp. 197-198 (2021)
Nayeon Kim
Department of AI Convergence
nayeunk1117@gm.gist.ac.kr

Curriculum Vitae

김나연 Nayeon Kim

Ph.D. Student

Affiliation & Contact

Affiliation
Gwangju Institute of Science and Technology (GIST), Gwangju, Republic of Korea

Research Interests

My research focuses on developing deep learning-based anomaly detection and continual learning frameworks for acoustic and fiber-optic sensing data in maritime and industrial environments.

Publications

International

  • Nayeon Kim, Minho Kim, Chanil Lee, Chanjun Chun, and Hong Kook Kim. “Novelty Detection in Underwater Acoustic Environments for Maritime Surveillance Using an Out-of-Distribution Detector for Neural Networks.” Sensors, 26(1) 37 (2025)

Domestic

  • 김나연, 박창수, 김홍국 “딥러닝 기반의 분포형 음향센서에서 DC Offset에 의한 이상 감지 성능의 개선 방법,.” 제33회 인공지능신호처리학술대회 논문집 (2023)
  • 정승훈, 권오현, 나형주, 마승희, 전슬기, 오현식, 이정혁, 김나연, 김홍국 “콜센터 데이터셋에서 음성인식 모델의 연속학습 연구,.” 2023년 한국통신학회 하계종합학술발표회 논문집, pp. 1594-1595 (2023)
  • 정승훈, 권오현, 나형주, 마승희, 전슬기, 오현식, 이정혁, 김나연, 김홍국 “한국어 음성인식을 위한 시퀀스 단위 연속학습 및 콜센터 적용,.” 2023 한국음성학회 봄 학술대회 논문집 (2023)
  • 김나연, 이건우, 박창수, 송민섭, 김홍국 “분포형 광섬유 진동 데이터를 활용한 인셉션 모델 기반 위험 상황 분류 연구,.” 2022년 한국통신학회 하계종합학술발표회 논문집, pp. 423-424 (2022)
  • 김나연, 이정혁, 박창수, 김홍국 “광섬유 OTDR 데이터의 이상 상태 감지를 위한 오토인코더 구조 비교 연구,.” 2021년 한국통신학회 추계종합학술발표회 논문집 (2021)
Seohyeon Shin
Department of Electrical Engineering and Computer Science
seohyeon@gm.gist.ac.kr

Curriculum Vitae

신서현 Seohyeon Shin

Ph.D. Student

Affiliation & Contact

Affiliation
Gwangju Institute of Science and Technology (GIST), Gwangju, Republic of Korea

Publications

International

  • Seohyeon Shin, HanJun Choi, Jun-Hyung Park, Hong Kook Kim, and Mansu Kim. “MolDA: Molecular Understanding and Generation via Large Language Diffusion Model.” CoRR (2026)
  • Seo Hyeon Shin, Kwang Myung Jeon, Nam Kyun Kim, Hong Kook Kim, Jeong Eun Lim, and Jinsoo Park. “Coordinate-based direction-of-arrival estimation method using distributed microphones.” ICCE 2018 (2018)

Integrated M.S.–Ph.D. Student

Dongkeon Park
Department of AI Convergence
dongkeon@gist.ac.kr

Curriculum Vitae

박동건 Donggun Park

Integrated M.S.–Ph.D. Student

Affiliation & Contact

Affiliation
Gwangju Institute of Science and Technology (GIST), Gwangju, Republic of Korea

Research Interests

My research interests focus on optimizing deep learning architectures, including LSTM, VAE, and residual networks, for diverse applications such as speech synthesis, seismic anomaly detection, and acoustic event classification.

Publications

International

  • Dongkeon Park, Yechan Yu, Dina Katabi, and Hong Kook Kim. “Adversarial Continual Learning to Transfer Self-Supervised Speech Representations for Voice Pathology Detection.” IEEE Signal Processing Letters, 30 932 (2023)
  • Dongkeon Park, Ji Won Kim, Kang Ryeol Kim, Do Hyun Lee, Hong Kook Kim “GIST-AiTeR Speaker Diarization System for VoxCeleb Speaker Recognition Challenge (VoxSRC) 2023,.” arXiv preprint, arXiv:2308.07788 (2023)
  • Dongkeon Park, Ji Won Kim, Kang Ryeol Kim, Do Hyun Lee, Hong Kook Kim “GIST-AiTeR Speaker Diarization System for VoxCeleb Speaker Recognition Challenge (VoxSRC) 2023.” arXiv:2308.07788 (2023)
  • Yechan Yu, Dongkeon Park, and Hong Kook Kim. “Auxiliary Loss of Transformer with Residual Connection for End-to-End Speaker Diarization.” ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (2022)
  • Dong Keon Park., Ye Chan Yu, Kyeong Wan Park, Ji Won Kim, Hong Kook Kim “GIST-AiTeR System for the Diarization Task of the 2022 VoxCeleb Speaker Recognition Challenge. arXiv preprint arXiv:2209.10357.” (2022)

Domestic

  • 임수환, 박동건, 김홍국 “한국어 FastSpeech2의 하이퍼파라미터 조절을 통한 최적화 연구,.” 2021년도 한국통신학회 하계종합발표회논문집, pp. 242-243 (2021)
  • 조건우, 박동건, 김홍국 “전리층 총 전자량 데이터에 적용한 LSTM 기반의 지진 이상현상 탐지,.” in 제1회 한국 인공지능 학술대회논문집 (Proc. 1st Korea Artificial Intelligence Conference), pp. 111-112 (2020)
  • 박동건, 김홍국 “퍼지 이산 학습기반의 변분 오토인코더,.” 제30회 신호처리합동학술대회논문집, pp. 24-25 (2020)
  • 김남균, 박동건, 김준호, 김홍국, 안충현 “청각장애인용 자막방송 서비스를 위한 연쇄잔차 신경망 기반 음향 사건 분류 기법,.” 2020년 한국방송미디어공학회 하계학술대회논문집, pp. 350-353 (2020)
Seunghun Jeong
Department of AI Convergence
zldzmfoq12@gm.gist.ac.kr

Curriculum Vitae

정승훈 Seunghun Jeong

Integrated M.S.–Ph.D. Student

Affiliation & Contact

Affiliation
Gwangju Institute of Science and Technology (GIST), Gwangju, Republic of Korea

Research Interests

My research focuses on developing advanced deep learning architectures, specifically Conformer and Graph Neural Networks, for robust speech recognition and spatiotemporal signal analysis in distributed acoustic sensing.

Publications

International

  • Seunghun Jeong, Huioon Kim, Young Ho Kim, Hyoyoung Jung, and Hong Kook Kim. “S2-DyGNN: A Spectro-Spatial Dynamic Graph Neural Network for Acoustic Event Classification in Distributed Acoustic Sensing.” Sensors, 26(14) 4417 (2026)
  • Seunghun Jeong, Huioon Kim, Huioon Kim, Young Ho Kim, Chang‐Soo Park, Hyoyoung Jung, Hong Kook Kim, and Hong Kook Kim. “Spatiotemporal Anomaly Detection in Distributed Acoustic Sensing Using a GraphDiffusion Model.” Sensors, 25(16) 5157 (2025)
  • Seunghun Jeong, Yejin Park, and Hong Kook Kim. “End-of-Sentence Token Modeling for Streaming Conformer-Based Korean Children’s Speech Recognition Applied to a Social Robot.” (2023)
  • Seunghun Jeong, and Hong Kook Kim. “Conformer-Based End-to-End Speech Recognition Using Grouped Convolution and Multi-Headed Linear Self-Attention with Headdrop.” (2023)

Domestic

  • 박예진, 정승훈, 김홍국 “레이블 스무딩과 레이어 정규화를 이용한 대화 화행 분류 성능 향상,.” 2023년 제4회 한국인공지능학술대회 논문집, pp. 392-393 (2023)
  • 정승훈, 권오현, 나형주, 마승희, 전슬기, 오현식, 이정혁, 김나연, 김홍국 “콜센터 데이터셋에서 음성인식 모델의 연속학습 연구,.” 2023년 한국통신학회 하계종합학술발표회 논문집, pp. 1594-1595 (2023)
  • 정승훈, 권오현, 나형주, 마승희, 전슬기, 오현식, 이정혁, 김나연, 김홍국 “한국어 음성인식을 위한 시퀀스 단위 연속학습 및 콜센터 적용,.” 2023 한국음성학회 봄 학술대회 논문집 (2023)
  • 정승훈, 김홍국 “트랜스포머 기반 한국어 인식 모델의 어린이 음성 성능 평가,.” 제32회 신호처리합동학술대회 논문집, pp. 251-253 (2022)
  • 마승희, 권오현, 나형주, 정승훈, 이정혁, 박현주, 김홍국 “긴 발화 콜센터 데이터에 적합한 한국어 음성 인식 모델 연구,.” 2022년 한국통신학회 하계종합학술발표회 논문집, pp. 1821-1822 (2022)
  • 정승훈, 김홍국 “전체 맥락과 깊이별 분리 합성곱을 이용한 Conformer 기반 음성 인식기의 순방향 모듈 개선,.” 2021년 한국음성학회 가을 학술대회 논문집, pp. 70 (2021)
Dohyun Lee Lab Leader
Department of AI Convergence
zerolee12@gm.gist.ac.kr

Curriculum Vitae

이도현 Dohyun Lee

Integrated M.S.–Ph.D. Student

Affiliation & Contact

Affiliation
Gwangju Institute of Science and Technology (GIST), Gwangju, Republic of Korea

Research Interests

My research focuses on enhancing speech processing technologies, specifically in text-to-speech synchronization for automated dubbing, audio source separation using LLM-based augmentation, and advanced speaker diarization and verification systems.

Publications

International

  • Changi Hong, Yoonah Song, Hwayoung Park, Chaewoon Bang, Dayeon Gu, Do Hyun Lee, and Hong Kook Kim. “PS-TTS: Phonetic Synchronization in Text-to-Speech for Achieving Natural Automated Dubbing.” CoRR (2026)
  • Do Hyun Lee, Yoonah Song, and Hong Kook Kim. “Performance Improvement of Language-Queried Audio Source Separation Based on Caption Augmentation From Large Language Models for DCASE Challenge 2024 Task 9.” CoRR (2024)
  • Dongkeon Park, Ji Won Kim, Kang Ryeol Kim, Do Hyun Lee, Hong Kook Kim “GIST-AiTeR Speaker Diarization System for VoxCeleb Speaker Recognition Challenge (VoxSRC) 2023,.” arXiv preprint, arXiv:2308.07788 (2023)
  • Dongkeon Park, Ji Won Kim, Kang Ryeol Kim, Do Hyun Lee, Hong Kook Kim “GIST-AiTeR Speaker Diarization System for VoxCeleb Speaker Recognition Challenge (VoxSRC) 2023.” arXiv:2308.07788 (2023)

Domestic

  • 이도현, 서연식, 송일훈, 김홍국 “화자 검증을 위한 트랜스포머 다중 특징 결합 풀링 모델,.” 제32회 신호처리합동학술대회 논문집, pp. 249-250 (2022)
Yoonah Song
Department of AI Convergence
yyaass0531@gm.gist.ac.kr

Curriculum Vitae

송윤아 Yoonah Song

Integrated M.S.–Ph.D. Student

Affiliation & Contact

Affiliation
Gwangju Institute of Science and Technology (GIST), Gwangju, Republic of Korea

Publications

International

  • Changi Hong, Yoonah Song, Hwayoung Park, Chaewoon Bang, Dayeon Gu, Do Hyun Lee, and Hong Kook Kim. “PS-TTS: Phonetic Synchronization in Text-to-Speech for Achieving Natural Automated Dubbing.” CoRR (2026)
  • Do Hyun Lee, Yoonah Song, and Hong Kook Kim. “Performance Improvement of Language-Queried Audio Source Separation Based on Caption Augmentation From Large Language Models for DCASE Challenge 2024 Task 9.” CoRR (2024)
  • Ji Won Kim, Sang Won Son, Yoonah Song, Hong Kook Kim, Il Hoon Song, Jeong Eun Lim “Semi-supervised learning-based sound event detection using frequency dynamic convolution with large kernel attention for DCASE Challenge 2023 Task 4,.” arXiv preprint, arXiv:2306.06461 (2023)
  • Ji Won Kim, Sang Won Son, Yoonah Song, Hong Kook Kim, Il Hoon Song, Jeong Eun Lim “Label filtering-based self-learning for sound event detection using frequency dynamic convolution with large kernel attention.” in Proc. of the 8th Detection and Classification of Acoustic Scenes and Events Workshop (DCASE) (2023)
  • Ji Won Kim, Sang Won Son, Yoonah Song, Hong Kook Kim, Il Hoon Song, and Jeong Eun Lim. “Semi-supervsied Learning-based Sound Event Detection using Freuqency Dynamic Convolution with Large Kernel Attention for DCASE Challenge 2023 Task 4.” CoRR (2023)
Jongyeon Park
Department of AI Convergence
jypark3737@gm.gist.ac.kr

Curriculum Vitae

박종연 Jongyeon Park

Integrated M.S.–Ph.D. Student

Affiliation & Contact

Affiliation
Gwangju Institute of Science and Technology (GIST), Gwangju, Republic of Korea

Research Interests

My research focuses on enhancing sound event detection and spatial semantic segmentation through domain-incremental learning, audio feature enrichment, and advanced classification architectures.

Publications

International

  • Jongyeon Park, Do-Hyeon Lim, Sang-won Park, Hong Kook Kim, Kyungdeuk Ko, Hyeongcheol Geum, and Jeong Eun Lim. “Domain-incremental audio classification using domain-specific experts and prototype classifier.” CoRR (2026)
  • Jongyeon Park, Joonhee Lee, Do-Hyeon Lim, Hong Kook Kim, Hyeongcheol Geum, and Jeong Eun Lim. “Performance improvement of spatial semantic segmentation with enriched audio features and agent-based error correction for DCASE 2025 Challenge Task 4.” CoRR (2025)
  • Sang Won Son, Jongyeon Park, Hong Kook Kim, Sulaiman Vesal, and Jeong Eun Lim. “Sound event detection based on auxiliary decoder and maximum probability aggregation for DCASE Challenge 2024 Task 4.” CoRR (2024)
Hwayoung Park
Department of AI Convergence
hwayoung_park@gm.gist.ac.kr

Curriculum Vitae

박화영 Hwayoung Park

Integrated M.S.–Ph.D. Student

Affiliation & Contact

Affiliation
Gwangju Institute of Science and Technology (GIST), Gwangju, Republic of Korea

Research Interests

My research focuses on enhancing the naturalness and quality of speech synthesis and voice conversion, with a particular emphasis on phonetic synchronization for automated dubbing and zero-shot voice cloning.

Publications

International

  • Changi Hong, Yoonah Song, Hwayoung Park, Chaewoon Bang, Dayeon Gu, Do Hyun Lee, and Hong Kook Kim. “PS-TTS: Phonetic Synchronization in Text-to-Speech for Achieving Natural Automated Dubbing.” CoRR (2026)
  • Hwa-Young Park, Chae-Woon Bang, Chanjun Chun, and Hong Kook Kim. “Pitch Encoder-Based Zero-Shot Voice Conversion for Improving Speech Quality.” ICAIIC 2025 (2025)
Hyeonwoo Park
Department of AI Convergence
hyeonwoo@gm.gist.ac.kr

Curriculum Vitae

박현우 Hyeonwoo Park

Integrated M.S.–Ph.D. Student, AI Convergence

Affiliation & Contact

Affiliation
Gwangju Institute of Science and Technology (GIST), Gwangju, Republic of Korea

Research Interests

Vision-grounded video-to-audio (sound effect) generation and audio-visual learning; sound event detection and acoustic scene analysis; edge / on-device AI and model lightweighting for real-time inference; speech processing (ASR and zero-shot TTS); sensor fusion; human-computer interaction.

Education

  • Sep. 2025 – PresentIntegrated M.S.–Ph.D. in AI Convergence — Gwangju Institute of Science and Technology (GIST)Advisor: Prof. Hong Kook Kim (AiTeR Lab)
  • Mar. 2021 – Aug. 2025B.S. in Computer Engineering — Chosun UniversityAdvisor: Prof. Chanjun Chun · GPA 3.865 / 4.5

Research Experience

  • Sep. 2025 – PresentGraduate Researcher (Integrated M.S.–Ph.D.) — AiTeR Lab, Dept. of AI Convergence, GISTVision-grounded video-to-audio (sound effect) generation, audio-visual learning, and AI-based audio/speech processing. Advisor: Prof. Hong Kook Kim.
  • Jan. 2022 – Aug. 2025Undergraduate Research Student — Advanced Multimedia Computing Lab., Chosun UniversitySound event detection, edge-AI model compression, and multimodal (audio + radar) danger-state recognition. Advisor: Prof. Chanjun Chun.

Publications

International Conferences

  • Hyeonwoo Park, Dayeon Ku, Jung Hyuk Lee, Hwayoung Park, Jongyeon Park, and Hong Kook Kim. “VisionSFX: Cross-Shot Consistent Video-to-Audio Generation with Depth-Aware Binaural Audio Rendering.” ICASSP 2026 Show & Tell, Barcelona, Spain, May 2026. (First author)
  • Hyeonwoo Park*, Dayeon Ku*, and Hong Kook Kim. “Listening to Motion in Space: Vision-Grounded Event-wise Video-to-Audio Generation and Rendering.” INTERSPEECH 2026 Show & Tell, Sydney, Australia, Sep.–Oct. 2026. (*Equal contribution / co-first author; to appear)

Domestic Conferences

  • Hyeonwoo Park, Heeran Jeong, and Chanjun Chun*. “Fine-Grained Bird Sound Event Detection Using Conformer.” Workshop on Convergent and Smart Media Systems, Korea Society of Smart Media, Jeju, Korea, Jan. 2023. (Oral presentation)

In Preparation

  • Hyeonwoo Park et al. “FastGuard: Edge-based Emergency Sound Detection for Public Toilets.” (Manuscript in progress)

Selected Projects

FastGuard – Edge-based Emergency Sound Detection for Public ToiletsMar. 2024 – Present

Lightweighting and field validation of a real-time AI emergency-bell system on edge devices — funded R&D, Chosun University Industry-Academic Cooperation Foundation (Nov.–Dec. 2024).

  • Built a CRNN-based sound event detector (6.52M params) optimized for low latency on a Jetson Orin Nano.
  • Applied FP16 quantization and TensorRT porting; used pooling and multi-threading to meet real-time constraints.

EarTalk – STT & Zero-Shot TTS for People with DysarthriaMay 2024 – Present

Speech assistive service that recognizes dysarthric speech and re-synthesizes it in a clear, personalized voice.

  • Developed the ASR/TTS pipeline: Whisper fine-tuning for STT and Coqui zero-shot TTS for voice synthesis.
  • Designed and led mobile app delivery (REST API, WebView) with Triton-based TensorRT/ONNX model serving.
  • Barrier-Free App Development Contest, Hyundai AutoEver (entry, Feb. 2025).

Sound Event Detection & Radar Peak Detection for Human Danger-State DetectionDec. 2023 – Aug. 2024

Complementary audio + radar sensing to detect elderly falls (60%+ occur at home) and enable fast, accurate reporting.

  • Trained a ResNet-18 sound event model on AI-Hub datasets; designed a radar peak-detection algorithm on TI data.
  • Deployed on Jetson Orin Nano with FP16 + TensorRT; built an Android app (Kotlin) for alerts.
  • Grand Prize (Director of IITP Award), ICT Smart Device Competition, Ministry of Science and ICT, Aug. 2024.

DABA (Dasi-Bada) – Recycled Plastic from Waste Fishing NetsAug. 2022 – Mar. 2024

Social-venture pipeline converting waste fishing nets into recycled PP/PE pellets and 3D-printing filament (collect → wash → extrude → cut).

  • Built the hardware (extruder, filament cutter, 3D printer) and an integrated control system for the production line.
  • Developed an app for end-of-job audio alerts and camera-linked task verification.
  • Awards: 3rd Place, Enactus National Competition (USA, Aug. 2023); Grand Prize, Hyundai Marine & Fire “Seed” Program (Jan. 2023); Gwangju Mayor’s Award, Social Venture Competition (Oct. 2023); Excellence Award, Gwangju Startup Festival (Nov. 2023).

Pet Cam & Pet ToyMay 2022 – Aug. 2022

Combined pet camera and interactive toy that detects a pet and triggers a motor to dispense treats.

  • Used a YOLOv3 pretrained detector on Raspberry Pi 4B; logged feeding times in MariaDB.
  • Built an Android app (camera, mic, speaker). Excellence Award, 88 Robot Day Share Challenge, Hanyang University ERICA (Aug. 2022).

Automatic Pill Dispenser for the ElderlyOct. 2021 – Jan. 2022

Device that sets medication times via OCR and recognizes utensils to ensure correct before/after-meal dosing.

  • Combined YOLOv3 object detection and Google OCR on a Raspberry Pi 4B; logged data in Firebase with calendar reminders.
  • Grand Prize, AI Robotics Convergence Idea Competition, Chosun University (Jan. 2022).

Honors & Awards

  • Grand Prize (Director of IITP Award), ICT Smart Device Competition, Ministry of Science and ICT — Aug. 2024
  • 3rd Place, Enactus National Competition, Enactus, United States — Aug. 2023
  • Grand Prize, Hyundai Marine & Fire Insurance “Seed” Program — Jan. 2023
  • Gwangju Metropolitan City Mayor’s Award, Social Venture Competition, Korea Social Enterprise Promotion Agency — Oct. 2023
  • Excellence Award, Gwangju Startup Festival, Gwangju Center for Creative Economy & Innovation — Nov. 2023
  • Excellence Award, 88 Robot Day Share Challenge (Adventure Design), Hanyang University ERICA — Aug. 2022
  • Grand Prize, AI Robotics Convergence Idea Competition, Chosun University — Jan. 2022

Work Experience

  • 2017 – 2021Staff, Mold Department — Taesung Industry Co., Ltd.Precision mold/machining work; CAD-based design.

Technical Skills

Languages
Python, Java, C/C++
Frameworks / Toolkits
PyTorch, Android (Android Studio, Kotlin)
Edge AI / Deployment
TensorRT porting, FP16 quantization, ONNX, Triton Inference Server, sub-10M-parameter model design
Hardware
NVIDIA Jetson (Nano, AGX Orin), Raspberry Pi, Arduino
Additional
AutoCAD design; technician certifications (CNC lathe/milling 2015, hydraulics 2016)

Languages

Korean (Native), English

 GitHub LinkedIn

M.S. Student

Junhee Lee
Department of Electrical Engineering and Computer Science
ljh13099@gm.gist.ac.kr
Sanghyun Choi
Department of Electrical Engineering and Computer Science
shchoiga@gm.gist.ac.kr
Jungwook Shim
Department of AI Convergence
tlawjddnr2@gm.gist.ac.kr
Dayeon Ku
Department of Electrical Engineering and Computer Science
dayeonku@gm.gist.ac.kr

Curriculum Vitae

구다연 Dayeon Ku

M.S. Student, Electrical Engineering and Computer Science

Affiliation & Contact

Affiliation
Gwangju Institute of Science and Technology (GIST), Gwangju, Republic of Korea

Research Interests

Multimodal machine learning with an emphasis on audio-visual learning; spoken language translation and its evaluation; text-to-speech; lip sync; speech and audio signal processing.

Education

  • Mar. 2025 – PresentM.S. in Electrical Engineering and Computer Science — Gwangju Institute of Science and Technology (GIST)
  • Sep. 2019 – Jun. 2023B.Eng. in Data and Systems Engineering — City University of Hong Kong
  • Sep. 2016 – Jun. 2019High School Diploma — International School of the Sacred Heart, Tokyo

Research & Industry Experience

  • Jul. 2025 – PresentAunionAI Co., Ltd.AI voice-technology startup (GIST faculty spin-off) specializing in automatic dubbing, audio localization, and speech synthesis.
  • Jun. – Nov. 2025Industry Project with Samsung ElectronicsDevelopment of AI dubbing solution technology for FAST channels.

Publications

International Conferences

  • Dayeon Ku, Hwayoung Park, and Hong Kook Kim. “Audiovisual CXMI: Scene-based Context Tagging for Spoken Language Translation Evaluation.” Proc. INTERSPEECH 2026, Sydney, Australia. (First and presenting author)
  • Hyeonwoo Park*, Dayeon Ku*, and Hong Kook Kim. “Listening to Motion in Space: Vision-Grounded Event-wise Video-to-Audio Generation and Rendering.” INTERSPEECH 2026 Show & Tell, Sydney, Australia. (Co-first author, equal contribution; presenting author)
  • Changi Hong*, Yoonah Song*, Hwayoung Park, Chaewoon Bang, Dayeon Ku, Do Hyun Lee, and Hong Kook Kim. “PS-TTS: Phonetic Synchronization in Text-to-Speech for Achieving Natural Automated Dubbing.” Proc. International Conference on Pattern Recognition (ICPR) 2026.
  • Hyeonwoo Park, Dayeon Ku, Jung Hyuk Lee, Hwayoung Park, Jongyeon Park, and Hong Kook Kim. “VisionSFX: Cross-Shot Consistent Video-to-Audio Generation with Depth-Aware Binaural Audio Rendering.” ICASSP 2026 Show & Tell, Barcelona, Spain. (Presenting author)

Domestic Conferences

  • Dayeon Ku, Hwayoung Park, and Hong Kook Kim. “Speech Pause Detection Using CTC Forced Alignment.” Proc. of the 2025 Summer Conference of the Korea Information and Communications Society (KICS), 2025. (First and presenting author; Best Paper Award)

Honors & Awards

  • Best Paper Award, 2025 Summer Conference of the Korea Information and Communications Society (KICS), 2025.

Languages

Korean (Native), English (Bilingual), Japanese (Intermediate)

 Scholar GitHub LinkedIn
Do-Hyeon Lim
Department of AI Convergence
do-hyeon@gm.gist.ac.kr

Curriculum Vitae

임도현 Do-Hyeon Lim

M.S. Student

Affiliation & Contact

Affiliation
Gwangju Institute of Science and Technology (GIST), Gwangju, Republic of Korea

Publications

International

  • Jongyeon Park, Do-Hyeon Lim, Sang-won Park, Hong Kook Kim, Kyungdeuk Ko, Hyeongcheol Geum, and Jeong Eun Lim. “Domain-incremental audio classification using domain-specific experts and prototype classifier.” CoRR (2026)
  • Jongyeon Park, Joonhee Lee, Do-Hyeon Lim, Hong Kook Kim, Hyeongcheol Geum, and Jeong Eun Lim. “Performance improvement of spatial semantic segmentation with enriched audio features and agent-based error correction for DCASE 2025 Challenge Task 4.” CoRR (2025)
Sangwon Park
Department of Electrical Engineering and Computer Science
kjun3088@gmail.com

Curriculum Vitae

박상원 Sangwon Park

M.S. Student

Affiliation & Contact

Affiliation
Gwangju Institute of Science and Technology (GIST), Gwangju, Republic of Korea

Research Interests

My research focuses on developing domain-incremental audio classification systems utilizing domain-specific experts and prototype-based classifiers.

Publications

International

  • Jongyeon Park, Do-Hyeon Lim, Sang-won Park, Hong Kook Kim, Kyungdeuk Ko, Hyeongcheol Geum, and Jeong Eun Lim. “Domain-incremental audio classification using domain-specific experts and prototype classifier.” CoRR (2026)
Jiung Park
Department of AI Convergence
zxwould4545@gmail.com

Undergraduate Intern

Yeongrim Ji
Department of Electrical Engineering and Computer Science
jyr79420ug@gm.gist.ac.kr