Research map / Image generation / Autoregressive and latent generation
Efficient Neural Music Generation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| increase DPD sampling steps from 5 to 20 (ablation) | MeLoDy · MusicCaps | FAD: 7.23 -> 5.41 | Helped |
| increase DPD sampling steps from 5 to 20 (ablation) | MeLoDy · MusicCaps | MCC: 0.49 -> 0.53 | Helped |
| increase DPD sampling steps from 5 to 20 (ablation) | MeLoDy · V100 GPU | speed: 181.1kHz -> 56.9kHz | Hurt |
| use MeLoDy instead of MusicLM | MeLoDy · music producer study | audio quality win proportion: 0.465 -> 0.535 | Helped |
| use MeLoDy instead of Noise2Music | MeLoDy · music producer study | audio quality win proportion: 0.436 -> 0.564 | Helped |
| use MeLoDy instead of MusicLM/Noise2Music | MeLoDy · music producer study | musicality and text correlation: 0.459 vs 0.541 (MLM musicality); 0.445 vs 0.555 (N2M) | No clear effect |
| use dual-path diffusion with LM semantic tokens instead of nested acoustic LMs | MeLoDy · sampling cost | NFEs: 4.32% and 0.41% of MusicLM NFEs for 10s and 30s | Helped |
| use proposed linear angle schedule instead of uniform schedule (ablation) | DPD · sampling with few steps | acoustic issues / MCC | Helped |
| use dual-path architecture instead of other LDM architectures (ablation) | DPD · training data subset | SNR improvement | Helped |
Closest papers by meaning
- Whole-Song Hierarchical Generation of Symbolic Music Using Cascaded Diffusion Models (ICLR 2024 · similarity 0.90)
- SongBloom: Coherent Song Generation via Interleaved Autoregressive Sketching and Diffusion Refinement (NeurIPS 2025 · similarity 0.90)
- Fast Timing-Conditioned Latent Audio Diffusion (ICML 2024 · similarity 0.90)
- Live Music Diffusion Models: Efficient Fine-Tuning and Post-Training of Interactive Diffusion Music Generators (NeurIPS 2026 · similarity 0.90)
- Presto! Distilling Steps and Layers for Accelerating Music Generation (ICLR 2025 · similarity 0.90)
- JEN-1 Composer: A Unified Framework for High-Fidelity Multi-Track Music Generation (AAAI 2025 · similarity 0.90)
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.