Research map / Image generation / Autoregressive and latent generation
Declarative Audio Editing with Audio Language Model
Zitong Lan, Yiduo Hao, Mingmin Zhao
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 ALM that decomposes declarative instruction into atomic steps (ablation) | SmartDJ · Declarative editing | FD: 23.6 -> 14.7 | Improved |
| add ALM that decomposes declarative instruction into atomic steps (ablation) | SmartDJ · Declarative editing | CLAP: 0.137 -> 0.238 | Improved |
| use audio-language ALM instead of captioner plus GPT-4o (ablation) | SmartDJ · Declarative editing | FD: 16.8 -> 14.7 | Improved |
| use remove-modify-add order instead of random order (ablation) | SmartDJ · Declarative editing | FD: 14.7 -> 14.7 | No clear effect |
| use remove-modify-add order instead of add-modify-remove (ablation) | SmartDJ · Declarative editing | FD: 14.9 -> 14.7 | No clear effect |
| SmartDJ editor instead of AudioSep for extraction | SmartDJ · Extract operation | FD: 27.1 -> 25.7 | Improved |
| SmartDJ editor instead of AudioSep for extraction | SmartDJ · Extract operation | LSD: 1.63 -> 1.71 | Hurt |
| add ALM to Audit baseline | Audit · Declarative editing | FD: 28.56 -> 21.50 | Improved |
| SmartDJ diffusion transformer editor instead of Audit | Audit · Single-step change sound direction | FD: 30.43 -> 9.45 | Improved |
| SmartDJ editor instead of Audit for volume edits | Audit · Single-step Volume | FD: 36.66 -> 9.19 | Improved |
| SmartDJ vs training-free SDEdit with ALM | SDEdit · Declarative editing | inference time: 301s -> 13.1s | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- Audit — AUDIT: Audio Editing by Following Instructions with Latent Diffusion Models (NeurIPS 2023) · better on 3 of 3 shared results
- SDEdit — SDEdit: Guided Image Synthesis and Editing with Stochastic Differential Equations (ICLR 2022) · better on 1 of 1 shared result
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
- SAO-Instruct: Free-form Audio Editing using Natural Language Instructions (NeurIPS 2025 · similarity 0.90)
- Prompt-guided Precise Audio Editing with Diffusion Models (ICML 2024 · similarity 0.89)
- ThinkSound: Chain-of-Thought Reasoning in Multimodal LLMs for Audio Generation and Editing (NeurIPS 2025 · similarity 0.88)
- AVENUE: Audio-Video Editing Understanding and Evaluation (NeurIPS 2026 · similarity 0.88)
- SoundBrush: Sound as a Brush for Visual Scene Editing (AAAI 2025 · similarity 0.87)
- Both Ears Wide Open: Towards Language-Driven Spatial Audio Generation (ICLR 2025 · similarity 0.86)
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.
