Papers › YourMT3+: Multi-instrument Music Transcription with Enhanced Transformer Architectures...

YourMT3+: Multi-instrument Music Transcription with Enhanced Transformer Architectures and Cross-dataset Stem Augmentation

5 Jul 2024arXiv:2407.04822archive 2025-07-28

Sungkyun Chang, Emmanouil Benetos, Holger Kirchhoff, Simon Dixon

Multi-instrument music transcription aims to convert polyphonic music recordings into musical scores assigned to each instrument. This task is challenging for modeling as it requires simultaneously identifying multiple instruments and transcribing their pitch and precise timing, and the lack of fully annotated data adds to the training difficulties. This paper introduces YourMT3+, a suite of models for enhanced multi-instrument music transcription based on the recent language token decoding approach of MT3. We enhance its encoder by adopting a hierarchical attention transformer in the time-frequency domain and integrating a mixture of experts. To address data limitations, we introduce a new multi-channel decoding method for training with incomplete annotations and propose intra- and cross-stem augmentation for dataset mixing. Our experiments demonstrate direct vocal transcription capabilities, eliminating the need for voice separation pre-processors. Benchmarks across ten public datasets show our models' competitiveness with, or superiority to, existing transcription models. Further testing on pop music recordings highlights the limitations of current models. Fully reproducible code and datasets are available with demos at \url{https://github.com/mimbres/YourMT3}.

PaperPDFCode

Code

mimbres/yourmt3 officialmentioned in papermentioned on GitHubGPL-3.0 report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Drum TranscriptionDrum Transcription in Music (DTM)Mixture-of-ExpertsMulti-Task LearningMulti-instrument Music TranscriptionMusic Transcription

Datasets

Introduced by this paper, per the archive.

YourMT3 Dataset

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Multi-instrument Music Transcription Slakh2100 YourMT3+ (YPTF.MoE+M) Multi F1 74.84 #1 of 5 Archive leaderboard report
Multi-instrument Music Transcription Slakh2100 MT3 Multi F1 62 #2 of 5 Archive leaderboard report
Multi-instrument Music Transcription Slakh2100 MT3 (colab) Multi F1 57.69 #4 of 5 Archive leaderboard report
Multi-instrument Music Transcription URMP YourMT3+ (YPTF.MoE+M) Multi F1 67.98 #1 of 3 Archive leaderboard report
Multi-instrument Music Transcription URMP MT3 Multi F1 59 #3 of 3 Archive leaderboard report
Music Transcription MAESTRO YourMT3+ (YPTF.MoE+M) noPS Onset F1 96.98 #4 of 9 Archive leaderboard report
Music Transcription MAESTRO YourMT3+ (YPTF.MoE+M) Onset F1 96.52 #6 of 9 Archive leaderboard report
Music Transcription MAPS YourMT3+ (YPTF.MoE+M, unseen) noPS Onset F1 88.73 #2 of 6 Archive leaderboard report
Music Transcription MAPS YourMT3+ (YPTF+S, unseen) Onset F1 88.37 #4 of 6 Archive leaderboard report
Music Transcription Slakh2100 YourMT3+ (YPTF.MoE+M) Onset F1 84.56 #1 of 6 Archive leaderboard report
Music Transcription Slakh2100 YourMT3+ (YPTF.MoE+M) note-level F-measure-no-offset (Fno) 0.8456 #1 of 6 Archive leaderboard report
Music Transcription Slakh2100 PerceiverTF Onset F1 81.9 #2 of 6 Archive leaderboard report
Music Transcription Slakh2100 PerceiverTF note-level F-measure-no-offset (Fno) 0.819 #2 of 6 Archive leaderboard report
Music Transcription Slakh2100 MT3 (colab) Onset F1 75.2 #3 of 6 Archive leaderboard report
Music Transcription Slakh2100 MT3 (colab) note-level F-measure-no-offset (Fno) 0.752 #3 of 6 Archive leaderboard report
Music Transcription URMP YourMT3+ (YPTF.MoE+M) Onset F1 81.79 #1 of 3 Archive leaderboard report
Music Transcription URMP MT3 Onset F1 77 #3 of 3 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

AttentionDropoutLinear LayerMoEMulti-Head AttentionPerceiver IORMSNormRotary EmbeddingsT5

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections