Text-to-score generation on 238 evaluation prompts
3.48Prompt AdherenceText2Score
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Text2Score2026.05 | 3.48 | 3.98 | 3.52 | 3.13 | 3.44 | |
| ComposerXLLM engine=GPT-5.12026.05 | 2.94 | 2.92 | 2.92 | 2.44 | 2.65 | |
| Midi-LLM2026.05 | 1.67 | 1.79 | 1.69 | 1.44 | 1.52 |