Multi-instrument transcription on Slakh2100 (test)
0.76F1 (Flat, Note-wise)[18]
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| [18]2022.06 | 0.76 | 0.57 | — | — | — | — | |
| pTS(s)Pretrained=Transcription module, Modules=Transcription and Separation, Merge method=Sum2022.06 | 0.584 | 0.262 | 0.612 | 0.285 | 0.506 | 0.247 | |
| ipTS(s)Pretrained=Transcription module, Modules=Transcription and Separation, Condition=Instrument recognition, Merge method=Sum2022.06 | 0.584 | 0.265 | — | — | — | — | |
| pTS(c)Pretrained=Transcription module, Modules=Transcription and Separation, Merge method=Concatenate2022.06 | 0.584 | 0.263 | 0.613 | 0.286 | 0.508 | 0.248 | |
| iTDescription=Transcription module only trained with instrument recognition conditions2022.06 | 0.58 | 0.254 | 0.597 | 0.272 | 0.487 | 0.232 | |
| TDescription=Transcription module only trained2022.06 | 0.579 | 0.251 | 0.597 | 0.272 | 0.487 | 0.232 | |
| TS(s)Modules=Transcription and Separation, Merge method=Sum2022.06 | 0.479 | 0.186 | 0.457 | 0.184 | 0.349 | 0.158 | |
| TS(c)Modules=Transcription and Separation, Merge method=Concatenate2022.06 | 0.477 | 0.186 | 0.46 | 0.18 | 0.355 | 0.162 | |
| [29]2022.06 | 0.266 | 0.134 | 0.115 | 0.063 | 0.043 | 0.019 |