Multi-instrument Automatic Music Transcription on Slakh 18 (test)
93.2BassYPTF.MoE+M (PS)
Evaluation results
| Method | Links | |||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| YPTF.MoE+M (PS)Vocabulary=MT3_MIDI_PLUS (MIDI plus singing)2024.07 | 93.2 | 74.96 | 67.7 | 82.27 | 73.48 | 88.84 | 74.72 | 82.22 | 75.44 | 84.19 | 45.57 | 90.05 | 84.56 | 70.7 | 90.05 | 74.84 | — | |
| PerceiverTF [25]Source=Original reported results2024.07 | 93 | 73.2 | 57.5 | 78.5 | 69.4 | 85.4 | 66.6 | 72.5 | 74.4 | 76.9 | 47.4 | 78.5 | 81.9 | — | 78.5 | — | 79.8 | |
| MT3 (colab [25])Source=Result reported by PerceiverTF author2024.07 | 90.6 | 43.3 | 34.31 | 73.2 | 36.3 | 78 | 28.2 | 44 | 55.1 | 40.9 | 23.4 | 77.3 | — | — | 77.3 | — | 74.3 | |
| MT3 (our colab)Source=Tested with our implementation on Colab notebook provided by MT3 author2024.07 | 71.03 | 28.67 | 34.31 | 65.9 | 30.14 | 70.87 | 40.6 | 19.41 | 47.02 | 29.51 | 20.02 | 83.85 | 75.2 | 56.78 | 83.85 | 57.16 | — | |
| MT3 [1]Source=Result reported by MT3 authors2024.07 | — | — | — | — | — | — | — | — | — | — | — | — | 76 | 57 | — | 62 | — |