Advancing Multi-Instrument Music Transcription: Results from the 2025 AMT Challenge

2026-03-29

AI Summary

This work addresses the core challenges in multi-instrument automatic music transcription (AMT), particularly polyphonic overlap and timbral variability, by organizing the 2025 AMT Challenge—a large-scale benchmark competition designed to evaluate state-of-the-art methods and advance the field. Using MT3 as the baseline system, the challenge attracted submissions from eight participating teams, two of which surpassed the baseline in both transcription accuracy and instrument identification. The study not only validates the efficacy of advanced deep learning architectures for multi-instrument AMT but also outlines key directions for future research, including broadening coverage across musical genres and strengthening instrument detection capabilities. Collectively, this effort establishes a robust benchmark and a clear pathway for ongoing progress in multi-instrument AMT.

Abstract

This paper presents the results of the 2025 Automatic Music Transcription (AMT) Challenge, an online competition to benchmark progress in multi-instrument transcription. Eight teams submitted valid solutions; two outperformed the baseline MT3 model. The results highlight both advances in transcription accuracy and the remaining difficulties in handling polyphony and timbre variation. We conclude with directions for future challenges: broader genre coverage and stronger emphasis on instrument detection.

Problem

• Automatic Music Transcription
• multi-instrument
• polyphony
• timbre variation
• instrument detection

Innovation

• multi-instrument transcription
• polyphony handling
• timbre variation
• MT3 baseline
• instrument detection