Public transcripts of Audrey Tang's speeches, interviews, and columns — roughly 2,000 markdown files from 2015 to the present. Published in the spirit of publishing before perishing: released as soon as possible, open to correction by anyone.
Served at archive.tw via sayit-hono.
File format
Each file is a standalone markdown document. The filename is the date and title:
2026-03-04-Ciudadanía-Digital.md 2026-03-04-數位公民.mdInside, transcripts follow a simple convention:
# 2026-03-04 數位公民
（speech content）
(Section Title)
（more content）
（question text）
（answer text） ```
# Date Title— document heading### Speaker:— speaker label (e.g.### Audrey Tang:or### 唐鳳：)> (Section Title)— section dividers### Question:/### 問：— questions in Q&A sections
Bilingual pairs
Many transcripts exist in both English and Traditional Chinese. The
.alternatesfile maps these pairs (tab-separated, one pair per line):2026-03-04-Ciudadanía-Digital.md 2026-03-04-數位公民.mdThe CI workflow reads this file and sends
alternate_filenameto the API so the two versions are linked on the site.How it works
git push *.md │ ▼ GitHub Actions workflow ├─ POST /api/upload_markdown → archive.tw (D1 + R2) └─ rebuild search index → deploy sayit-hono to Cloudflare WorkersOn every push that touches a
.mdfile, the workflow in.github/workflows/upload-markdown-on-change.yml:- Detects added, modified, and deleted markdown files
- Uploads (or deletes) them via the archive.tw API, including
alternate_filenamefor bilingual pairs - Rebuilds the Pagefind search index and redeploys the site
Contributing
If you spot an error in any transcript, open a pull request. The content is CC0 — public domain.
License