ML Research Wiki / Benchmarks / Music Transcription / MAESTRO

MAESTRO

Music Transcription Benchmark

Performance Over Time

📊 Showing 6 results | 📏 Metric: Onset F1

Top Performing Models

Rank Model Paper Onset F1 Date Code
1 Transkun V2 (SemiCRF) Scoring Time Intervals using Non-Hierarchical Transformer For Automatic Piano Transcription 98.32 2024-04-15 📦 yujia-yan/transkun 📦 yujia-yan/skipping-the-frame-level
2 hFT-Transformer Automatic Piano Transcription with Hierarchical Frequency-Time Transformer 97.44 2023-07-10 📦 sony/hft-transformer
3 YourMT3+ (YPTF.MoE+M) noPS 📚 YourMT3+: Multi-instrument Music Transcription with Enhanced Transformer Architectures and Cross-dataset Stem Augmentation 96.98 2024-07-05 📦 mimbres/yourmt3
4 YourMT3+ (YPTF.MoE+M) 📚 YourMT3+: Multi-instrument Music Transcription with Enhanced Transformer Architectures and Cross-dataset Stem Augmentation 96.52 2024-07-05 📦 mimbres/yourmt3
5 MT3 (single dataset) MT3: Multi-Task Multitrack Music Transcription 88.00 2021-11-04 📦 magenta/mt3 📦 mimbres/yourmt3 📦 kunato/mt3-pytorch
6 MT3 (multi dataset) 📚 MT3: Multi-Task Multitrack Music Transcription 86.00 2021-11-04 📦 magenta/mt3 📦 mimbres/yourmt3 📦 kunato/mt3-pytorch

All Papers (6)

yujia-yan/transkun yujia-yan/skipping-the-frame-level
sony/hft-transformer
mimbres/yourmt3
mimbres/yourmt3
magenta/mt3 mimbres/yourmt3 kunato/mt3-pytorch
magenta/mt3 mimbres/yourmt3 kunato/mt3-pytorch