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

Benchmark Model Rank Results
multi-instrument-music-transcription-onYourMT3+ (YPTF.MoE+M)#1Multi F1: 74.84
multi-instrument-music-transcription-onMT3#2Multi F1: 62
multi-instrument-music-transcription-onMT3 (colab)#4Multi F1: 57.69
multi-instrument-music-transcription-on-urmpYourMT3+ (YPTF.MoE+M)#1Multi F1: 67.98
multi-instrument-music-transcription-on-urmpMT3#3Multi F1: 59
music-transcription-on-maestroYourMT3+ (YPTF.MoE+M) noPS#3Onset F1: 96.98
music-transcription-on-maestroYourMT3+ (YPTF.MoE+M)#5Onset F1: 96.52
music-transcription-on-mapsYourMT3+ (YPTF.MoE+M, unseen) noPS#2Onset F1: 88.73
music-transcription-on-mapsYourMT3+ (YPTF+S, unseen)#3Onset F1: 88.37
music-transcription-on-slakh2100YourMT3+ (YPTF.MoE+M)#1note-level F-measure-no-offset (Fno): 0.8456Onset F1: 84.56
music-transcription-on-slakh2100PerceiverTF#2note-level F-measure-no-offset (Fno): 0.819Onset F1: 81.9
music-transcription-on-slakh2100MT3 (colab)#3note-level F-measure-no-offset (Fno): 0.752Onset F1: 75.2
music-transcription-on-urmpYourMT3+ (YPTF.MoE+M)#1Onset F1: 81.79
music-transcription-on-urmpMT3#3Onset F1: 77