Research map / Image generation / Text-to-image diffusion
MIDILM: A Dual-Path Model for Controllable Text-to-MIDI Generation
Shuyu Li, Dooho Choi, Yunsick Sung
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add modality isolation (dual-path) (ablation) | MIDILM · MidiCaps test | TB: 0.8694 -> 0.9688 | Improved |
| add MIDI-side MoE routing (ablation) | MIDILM · MidiCaps test | TB: 0.9132 -> 0.9688 | Improved |
| add both modality isolation and MoE (ablation) | MIDILM · MidiCaps test | TB: 0.8269 -> 0.9688 | Improved |
| add modality isolation (ablation) | MIDILM · MidiCaps test | CK: 0.4159 -> 0.4688 | Improved |
| add MIDI-side MoE routing (ablation) | MIDILM · MidiCaps test | CK: 0.4426 -> 0.4688 | Improved |
| add modality isolation (ablation) | MIDILM · MidiCaps test | CLaMP 3: 0.1405 -> 0.1564 | Improved |
| add MIDI-side MoE routing (ablation) | MIDILM · MidiCaps test | CLAP: 0.3054 -> 0.3127 | Improved |
| add both components (ablation) | MIDILM · MidiCaps test | CR: 2.2139 -> 2.4617 | Improved |
| MIDILM vs Text2MIDI-InferAlign on free-form texts | MIDILM · Free-form texts | CLAP: 0.1847 -> 0.1881 | Improved |
| MIDILM vs Text2MIDI-InferAlign on free-form texts | MIDILM · Free-form texts | CR: 2.1414 -> 3.1102 | Improved |
| MIDILM vs Text2MIDI-InferAlign in listening test | MIDILM · MidiCaps test | MOS: 2.4861 -> 3.2306 | Improved |
| MIDILM vs Text2MIDI-InferAlign in listening test | MIDILM · Free-form texts | MOS: 3.2965 -> 3.3417 | Improved |
Closest papers by meaning
- Text2midi: Generating Symbolic Music from Captions (AAAI 2025 · similarity 0.90)
- MIDI-GPT: A Controllable Generative Model for Computer-Assisted Multitrack Music Composition (AAAI 2025 · similarity 0.89)
- CSL-L2M: Controllable Song-Level Lyric-to-Melody Generation Based on Conditional Transformer with Fine-Grained Lyric and Musical Controls (AAAI 2025 · similarity 0.89)
- MuseControlLite: Multifunctional Music Generation with Lightweight Conditioners (ICML 2025 · similarity 0.89)
- MuPT: A Generative Symbolic Music Pretrained Transformer (ICLR 2025 · similarity 0.88)
- CMI-RewardBench: Evaluating Music Reward Models with Compositional Multimodal Instruction (ICML 2026 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
