A curated, bilingual research index covering audio effects, differentiable processing, representation learning, staged performance rendering, production programs, automatic mixing, mastering, spatial audio, and evaluation.
Browse the index · Submit a resource · Report an issue
185 papers · 108 verified projects · 24 datasets · 3 reference collections
| Collection | Included information |
|---|---|
| Papers | Formal venues, paper and DOI links, bilingual summaries, production stages, output types, method classifications, and associated open resources |
| Projects | Source, checkpoint, inference, training, dataset, license, taxonomy, related-paper evidence, and last-verification dates |
| Datasets | Contents, access conditions, data licenses, scale, and evidence-backed links to papers and projects that use each dataset |
| Reference resources | Field bibliographies and research guides kept separate from runnable implementations |
| Weekly update | A short bilingual overview and up to three linked highlights from the latest accepted catalogue update |
The index is organized by production task rather than publication year. It currently covers:
- audio effects modeling, estimation, control, transfer, and removal;
- differentiable DSP, effect chains, and processing graphs;
- effect and production-style representation learning;
- editable symbolic performance and performance-to-audio rendering;
- production programs, processing graphs, DAW interaction, and agentic editing;
- automatic, reference-guided, and controllable mixing;
- mastering and remastering;
- benchmarks, metrics, datasets, and reproducibility tools;
- intelligent spatial generation, upmixing, rendering, HRTF personalization, and production evaluation.
- Paper titles, authors, dates, and links are anchored to primary publication sources where available.
- Formal venues are shown only when publication evidence can be verified.
- Project source, checkpoint, dataset, taxonomy, and license claims are recorded with first-party repository or official-documentation evidence.
- The weekly paper scout proposes reviewable changes, including a plain-language homepage summary; AI-screened candidates and generated copy are never published directly.
- AI Highlight and year-normalized citation impact remain separate, documented signals rather than a combined score.
- Every public entry remains open to correction through a structured issue or pull request.
The quickest route is to open a structured request:
Please include a primary paper or project URL and enough evidence to verify the requested change. See CONTRIBUTING.md before editing catalogue data directly.
The catalogue is available as version-controlled JSON:
These files drive the public website and can also be used for research tooling or downstream analysis under the repository license.
See DATA.md for endpoint URLs, schema versions, field semantics, and migration guidance. Formal contracts are available for projects v3, datasets v1, and weekly updates v1.
The weekly paper scout searches arXiv, verifies related GitHub resources, screens candidates with a configured language model, and writes complete paper records with short names, bilingual summaries, production stages, output types, classifications, AI assessments, publication metadata, and resource-review results. Verified source repositories create linked project records; documentation-only pages are kept distinct from source code. The workflow then refreshes DOI, venue, and citation metadata, writes a concise bilingual update, and validates every relation before publishing to main. A transient GitHub search failure is recorded as a deferred review rather than "not found" and does not block the remaining update.
See automation/README.md for configuration and review rules.
- Establish a bilingual index for papers, projects, datasets, and reference collections.
- Link datasets to verified papers and open-source projects that use them.
- Audit historical coverage across effect modeling, parameter estimation, representation learning, mixing, and mastering.
- Add a focused spatial-audio production taxonomy and initial evidence-backed catalogue.
- Recheck PEACE (joint DSP-effects code and audio embeddings) when a verifiable paper or preprint becomes public.
- Verify installation, inference, training, and checkpoint paths for high-impact repositories.
- Add structured benchmark metadata for comparing tasks, datasets, metrics, and reported results.
- Expand automated checks for formal venues, licenses, stale repositories, and broken links.
- Run a validated weekly paper scout with optional pull-request review mode.
- Add a lightweight contributor review process for recurring community submissions.
- Publish versioned catalogue snapshots and document a stable downstream data interface.
The site code, automation, and original catalogue content in this repository are released under the MIT License. Linked papers, codebases, models, datasets, and trademarks remain subject to their respective owners' terms.