Text-conditioned music generation on Song Describer Dataset
29FDLMDM (ED)+AF
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| LMDM (ED)+AFBlocks=21, Sampler=Ping-Pong, w/Priming?=true2026.05 | 29 | 0.35 | 32 | 0.03 | 8 | |
| LMDM (ED)Blocks=21, Sampler=Euler, w/Priming?=true2026.05 | 35.35 | 0.62 | 23 | 0.11 | 50 | |
| LMDM (BC)+AFBlocks=21, Sampler=Ping-Pong, w/Priming?=true2026.05 | 35.45 | 0.53 | 23 | 0.02 | 2 | |
| LMDM (ED)+AFBlocks=21, Sampler=Ping-Pong, w/Priming?=false2026.05 | 35.88 | 0.74 | 29 | 0.03 | 8 | |
| LMDM (BC)Blocks=21, Sampler=Euler, w/Priming?=true2026.05 | 47.13 | 0.74 | 24 | 0.17 | 50 | |
| LMDM (BC)+AFBlocks=21, Sampler=Ping-Pong, w/Priming?=false2026.05 | 47.26 | 0.91 | 23 | 0.02 | 2 | |
| LMDM (ED)Blocks=21, Sampler=Euler, w/Priming?=false2026.05 | 61.06 | 1.14 | 20 | 0.11 | 50 | |
| LMDM (BC)Blocks=21, Sampler=Euler, w/Priming?=false2026.05 | 64.87 | 1.2 | 20 | 0.17 | 50 | |
| Magenta RealTimeBlocks=24, w/Priming?=false2026.05 | 72.14 | 0.47 | 35 | 4 | 800 | |
| Stable Audio OpenBlocks=1, Sampler=DPM++, w/Priming?=false2026.05 | 96.51 | 0.55 | 41 | 10.35 | 100 | |
| MusicGen-LargeBlocks=1, w/Priming?=false2026.05 | 190.47 | 0.52 | 31 | 10.81 | 2.4 |