Research map / Image generation / Editing and style transfer
SteerMusic: Enhanced Musical Consistency for Zero-shot Text-Guided and Personalized Music Editing
Xinlei Niu, Kin Wai Cheuk, Jing Zhang, Naoki Murata, Chieh-Hsin Lai, Michele Mancusi, Woosung Choi, Giorgio Fabbro, Wei-Hsiang Liao, Charles Patrick Martin, Yuki Mitsufuji
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 |
|---|---|---|---|
| edit in data space with delta denoising score instead of forward-backward diffusion (vs ZETA) | SteerMusic · ZoME-Bench | CQT1-PCC: 0.293 -> 0.480 | Improved |
| delta-score data-space editing vs ZETA on source consistency | SteerMusic · ZoME-Bench | FAD_Vggish: FAD_Vggish 3.380 -> 2.426 | Improved |
| SteerMusic vs DDIM on edit fidelity | SteerMusic · ZoME-Bench | CLAP: CLAP 0.264 (DDIM) vs 0.259 | Hurt |
| SteerMusic vs best baseline on listener preference | SteerMusic · ZoME-Bench | MOS: MOS-P 2.92; MOS-T 2.50 | Improved |
| SteerMusic on long-form MusicDelta clips vs ZETA | SteerMusic · MusicDelta | CQT1-PCC: CQT1-PCC 0.296 -> 0.351 | Improved |
| personalized delta score with data-space editing vs DreamSound | SteerMusic+ · ZoME-Bench personalized | FAD_CLAP: FAD_CLAP 0.902 -> 0.362 | Improved |
| SteerMusic+ vs DreamSound on concept alignment | SteerMusic+ · ZoME-Bench personalized | CDPAM: CDPAM 0.609 -> 0.593 | Improved |
| SteerMusic+ vs DreamSound on long-form MusicDelta | SteerMusic+ · MusicDelta | FAD_Vggish: FAD_Vggish 8.972 -> 5.506; FAD_CLAP 0.847 -> 0.666 | Mixed |
| add distribution shift regularization L_shift to personalized delta score (ablation) | SteerMusic+ · bouzouki concept | CDPAM | No clear effect |
| add personalized contrastive loss PCon (ablation) | SteerMusic+ · bouzouki concept | CDPAM | Improved |
| increase regularization weight lambda to positive values (ablation) | SteerMusic+ · bouzouki concept | CDPAM and CQT1-PCC | Mixed |
| more PDM fine-tuning steps (50 vs 200) (ablation) | SteerMusic+ and DreamSound · personalized editing | CDPAM and CQT1-PCC | Mixed |
Later papers that compared with this method
1 later accepted paper reports SteerMusic as a baseline in their results tables; 0 of them report a better number on most of the results they share with it.
- Polyphonia: Zero-Shot Timbre Transfer in Polyphonic Music with Acoustic-Informed Attention Calibration (ICML 2026 · better on 2 of 6 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Zero-Shot Unsupervised and Text-Based Audio Editing Using DDPM Inversion (ICML 2024 · similarity 0.89)
- SongEditor: Adapting Zero-Shot Song Generation Language Model as a Multi-Task Editor (AAAI 2025 · similarity 0.87)
- Music Style Transfer with Time-Varying Inversion of Diffusion Models (AAAI 2024 · similarity 0.85)
- Efficient Fine-Grained Guidance for Diffusion Model Based Symbolic Music Generation (ICML 2025 · similarity 0.84)
- MeLFusion: Synthesizing Music from Image and Language Cues using Diffusion Models (CVPR 2024 · similarity 0.84)
- DITTO: Diffusion Inference-Time T-Optimization for Music Generation (ICML 2024 · similarity 0.84)
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.
