Chau Hoang

Chau Hoang

I am a Ph.D. student in Computer Science at the University of Maryland, College Park, advised by Prof. Ming C. Lin and Prof. Dinesh Manocha. I am a member of GAMMA Lab, where I work on grounded auditory intelligence for long-horizon spoken interaction. My research focuses on building online audio-language agents that can maintain and revise an evidence-grounded beliefs under partial observation as evidence is revealed incrementally, and use its uncertainty to decide when to continue listening, clarify, defer, or act.

Previously, I was an AI Engineer at Viettel AI, working on simultaneous speech translation, robust automatic speech recognition, and speech processing systems for the company’s products. Before that, I conducted research at Sensor Lab, Hanoi University of Science and Technology (HUST), primarily on speech enhancement and efficient speech representation learning. My work has resulted in publications in IEEE/ACM TASLP, Applied Acoustics, and Measurement. I received my M.S. and B.S. degrees in Control Engineering and Automation from HUST in 2024 and 2022, respectively.

Research Interests

  • Audio, Speech, and Signal Processing

  • Long-horizon Conversational Agents

  • Multi-modal Learning

📝 Publications

Selected Publications

IEEE/ACM Transactions on Audio, Speech, and Language Processing
sym

A Novel Approach to Multi-Channel Speech Enhancement Based on Graph Neural Networks

Ngoc Chau Hoang, Tien Dat Bui, Huu Binh Nguyen, Thanh Thi Hien Duong, Quoc Cuong Nguyen

  • Proposed a dynamic and explicit information aggregation method to model latent speech representation more effectively in real-time multi-channel speech enhancement.
Applied Acoustics
sym

Multi-Stage Temporal Representation Learning via Global and Local Perspectives for Real-time Speech Enhancement

Ngoc Chau Hoang, Thi Nhat Linh Nguyen, Tuan Kiet Doan, Quoc Cuong Nguyen

  • Introduced useful inductive bias to global modeling operations (self-attention and graph convolution).
Measurement
sym

Advancing Robust Human Activity Recognition via Informative mmWave Radar Characteristics and A Lightweight Spatio-Spectro-Temporal Network

Van Ngoc Dang, Ngoc Chau Hoang, Quoc Cuong Nguyen, Minh Thuy Le

  • Developed a mmWave radar-based human activity recognition system that can adaptively model complex spatial dependencies among receivers, enabling it to distinguish the direction of body parts’ movements. This is especially beneficial for recognizing activities performed along arbitrary trajectories.
IEEE Sensors Journal
sym

Deep Learning-based Human Activity Recognition with FMCW Radar: A Review

Van Ngoc Dang, Ngoc Chau Hoang, Minh Thuy Le, Kien Nguyen, Quoc Cuong Nguyen

  • Reviewed the state of the art on deep learning-based human activity recognition using FMCW radar, including radar signal processing techniques, model architectures, and existing datasets.
IEEE Transactions on Audio, Speech, and Language Processing
sym

Speaker-conditioned U-shaped Diarization with Speaker Extraction-guided Enhancement

Ngoc Thuan Tran, Ngoc Chau Hoang, Quoc Cuong Nguyen

  • Introduced hierarchical fusion of speech separation features and speaker embeddings to bridge the resolution gap between two tasks, aligning the separation task with the inherent nature of diarization.

Other Publications

📖 Education

  • 2026.08 - Present, Ph.D., University of Maryland, College Park, MD, USA.
  • 2022.12 - 2024.12, Master, Hanoi University of Science and Technology, Hanoi, Vietnam.
  • 2018.08 - 2022.09, Undergraduate, Hanoi University of Science and Technology, Hanoi, Vietnam.

🎖 Awards and Service

  • 2024.02, Graduate Scholarship: Awarded to outstanding graduate students.
  • 2020 - 2021, Study Encouragement Scholarship : Awarded to top 2% excellent students out of a total of more than 30,000 undergraduate students by Hanoi University of Science and Technology. (Fall’20, Spring’21, Fall’21).
  • Reviewer for Journal: Machine Learning: Science and Technology (MLST)

💻 Work Experience

Visitor Geography