Speech Processing Lab

Our research focuses on speech and audio intelligence, with an emphasis on robust representation learning and clinical applications.

• Speech and Audio Foundation Models: Self-supervised learning, representation learning, and model adaptation

• Automatic Speech Recognition: End-to-end ASR, speech understanding, and error analysis in geriatrics

• Respiratory Sound Analysis: Biomedical acoustics, respiratory sound classification, and acoustic biomarkers

• Speech-Based Digital Biomarkers: Neurodegenerative disorders, psychiatric conditions, and speech disorders

• Multimodal Biomedical AI: Speech, physiological signals, images, and eye-tracking analysis

2026. 10. "Acquisition and Demographic Factors Shape Pathology-Related Variability in Respiratory Sound Analysis" has been accepted at Frontiers in Physiology (IF 4.3).

2026. 10. "Automatic Speech Recognition Error Patterns for Differentiating Clinically Diagnosed Dementia, Parkinson's Disease, and Healthy Aging: Cross-Sectional Association Study" has been accepted at JMIR Aging (IF 4.9).

2026. 09. "Integrating patient narratives and clinical indicators for monitoring suicidal ideation in adolescents with major depressive disorder: a human-in-the-loop validation study of LLM-assisted journaling" has been accepted at BMC Psychiatry (IF 4.6). [webpage]

2026. 09. "Language Orthogonalization of Self-Supervised Speech Representations for Cross-lingual Parkinson's Detection" has been accepted at IEEE SLT 2026. [pdf]

2026. 09. Our lab has been selected for the Core Research Program by the National Research Foundation of Korea for 2026-2028 (핵심연구 기본연구A).

2026. 08. "DSM-Guided Large Language Model Reasoning for Depression and PTSD Assessment From Psychiatric Interview Transcripts" has been accepted at IEEE Access 2026 (IF 4.2). [webpage]

2026. 06. "Lung-SRAD: Spectral-Aware Regularized Audio DASS with Dual-Axis Patch-Mix Contrastive Learning for Respiratory Sound Classification" has been accepted at Interspeech 2026. [webpage]

2026. 06. "Quality Adaptive Angular Margin Learning for Respiratory Sound Classification" has been accepted at Interspeech 2026. [webpage]

2026. 06. "A Hierarchical Feature Engineering Framework for Automated Classification of Phonotraumatic and Non-Phonotraumatic Vocal Hyperfunction" has been accepted at Interspeech 2026. [webpage]

2026. 05. "JODAL: Joint Domain Adversarial Learning for TTS-Augmented Automatic Speech Recognition" has been published at Mathematics 2026 (IF 2.2 ). [webpage]

2026. 05. Our lab has been selected for the 2026 R&D CareerLab Fellowship by the Korea Institute of Human Resources Development in Science and Technology (KIRD) (R&D 커리어랩 펠로우십 - 미래 의사과학자 챌린지 트랙).

2026. 04. "Meta Ensemble Learning with Diverse Data Splits for Improved Respiratory Sound Classification" has been accepted at IEEE EMBC 2026. [pdf]

2026. 04. "Recording Conditions Dominate Acoustic Feature Variability in Respiratory Sound Analysis" has been accepted at IEEE EMBC 2026 (One-Page Paper).

2026. 01. "Deep neural network-based analysis of voice biomarkers for monitoring treatment response in adolescent major depressive disorder" has been published at Communications Medicine 2026 (IF 6.3). [webpage]

2026. 01. "Understanding Frechet Speech Distance for Synthetic Speech Quality Evaluation" has been accepted at ICASSP 2026. [webpage]

2026. 01. "Empowering Multimodal Respiratory Sound Classification with Counterfactual Adversarial Debiasing for Out-of-Distribution Robustness" has been accepted at ICASSP 2026. [webpage]

Notice

Graduate Student Recruitment (B.S. / M.S. / Ph.D)

We are seeking highly motivated B.S. (Korean only; medical students are also welcome), M.S. (Korean only), Ph.D. students with strong technical backgrounds and research experience in speech processing, audio AI, and deep learning. We particularly welcome applicants with prior research experience in one or more of the following areas:

• Speech and Audio Processing: ASR, speech understanding, self-supervised learning, and speech/audio foundation models

• Biomedical Signal Analysis: Respiratory sounds, neurological and psychiatric speech analysis, and physiological signals (QEEG, Vital signs, ECG, PPG)

• Multimodal AI: Speech, audio, and physiological signal integration

Graduate student applicants should have strong Python and PyTorch programming skills as well as experience in implementing and evaluating deep learning models. For Ph.D. applicants, at least two papers are required. Prior research experience, publications, or participation in relevant research projects will be highly valued. Interested applicants should send their CV, academic transcripts, and a brief description of their research experience to kaen2891@wku.ac.kr.