Ting-Yu Pan
AI Music · Machine Learning · Signal Processing

Sound, Code, &
Research

Designing AI music systems that enhance creativity, support performance, and connect human expression with machine learning.

Machine Learning Audio Computing Generative Models AI Music Human–AI Co-Creation Music Education Tech

Research Interests

My work focuses on building impactful AI music systems grounded in real musical practice. I explore generative music models, machine-learning–based audio modeling, human–AI co-creativity, and interactive tools for rehearsal, performance, and learning.

I aim to create systems that enhance musicians’ creativity and musical understanding—never replacing artistry, but supporting expressive, collaborative workflows.

Generative models for expressive musical structure
Machine learning for audio analysis and source modeling
Human–AI co-creativity in rehearsal and composition
Intelligent tools for music learning & performance support
Dataset design & evaluation for AI music research

Research

AcaMate: Supporting Novice A Cappella Singers in Iterative Individual Practice

ACM UIST 2025 (Accepted)

An AI-assisted rehearsal tool that visualizes pitch, rhythm, and dynamics to help novice a cappella singers practice more effectively with group recordings and receive structured feedback.

Paper

Curating an A Cappella Dataset for Source Separation

ISMIR 2025 Late-Breaking Demo (Accepted)

A studio-quality, multilingual a cappella dataset and experimental pipeline for two-step source separation, focusing on vocal percussion isolation and SATB separation for more realistic rehearsal and production scenarios.

Paper

ACappellaSet: A Multilingual A Cappella Dataset for Source Separation and AI-assisted Rehearsal Tools

NeurIPS 2025 — AI for Music Workshop (Accepted)

A 55-song multilingual a cappella dataset (SATB + vocal percussion) recorded by professional ensembles, designed for training and evaluating models in source separation, rehearsal support, and AI music applications.

Paper

Technical Projects

AcaMate / ACappellaSet Web Demo

A browser-based demo showcasing AI-assisted rehearsal features powered by ACappellaSet, including SATB visualization, vocal percussion extraction, and practice feedback tools for a cappella singers. Built in collaboration with Kexin Phyllis Ju and Prof. Hao-Wen Dong.

A Cappella Source Separation Human–AI Music

Curriculum Vitae

I’m currently refining my full CV. If you’d like the most up-to-date version or more details about my work, please feel free to reach out via email.

© 2025 Ting-Yu Pan.