Two thousand speech papers, one very long markdown table

What it does This repository catalogs accepted papers from ICASSP 2023 and 2024—roughly 2,700 entries for 2024 alone, with about 79% currently indexed—and lays them out in extensive markdown tables sorted by technical track. Each listing links to the paper and, when available, its external code repository. Behind the scenes, GitHub Actions parse these tables into JSON and mirror the dataset to a Hugging Face app for easier browsing.

The interesting bit It treats a conference proceedings like a content-management system that needs ETL. Instead of a static reading list, the maintainers have built automation to keep markdown tables synchronized with a structured JSON backend and a web frontend, effectively turning the repo into the database layer for a lightweight paper explorer.

Key highlights

• Covers the full ICASSP 2023 and 2024 technical programs, spanning speech recognition, audio generation, denoising, multimodal learning, and signal restoration.
• Links to external code repositories are included when authors have published them.
• GitHub Actions workflows automatically parse the markdown tables and generate structured JSON for downstream consumption.
• A companion Hugging Face app (NewEraAI-Papers) provides a searchable interface.
• Part of a larger family of conference indexes maintained by the same author, including CVPR, INTERSPEECH, and AAAI.

Caveats

• The 2024 collection is a work in progress: the README shows 79% completion, so roughly one in five papers is still missing.
• The repository is primarily markdown tables and automation glue; it is an index, not a framework or library.

Verdict Worth bookmarking if you are doing a literature review in speech or audio processing and want a centralized jumping-off point with code links. Skip it if you are hunting for original implementations or ready-to-run training scripts.