Super Minutes

A privacy-respecting web app that turns a meeting recording into a structured minutes report:

1. The user picks an audio or video file locally.
2. If it's a video, audio is extracted in the browser with ffmpeg.wasm — the video itself never leaves the device.
3. The audio is sent to 榛果繽紛樂's OpenAI-compatible Whisper Gateway (e.g. whisper-gateway:5000) with speaker diarization on.
4. The diarized transcript is fed to a local Ollama instance (default model gemma4:e4b) to draft a structured minutes report: 會議紀要 (summary) · 主要結論 (conclusions) · 議題與要點 (topics & points) · 待辦行動項 (action items).
5. The user can edit the transcript inline, regenerate the report, switch between Traditional / Simplified Chinese (in-browser via OpenCC), then export the package as Markdown, DOCX, or PDF.

Built with Next.js 15 (App Router), TypeScript, Tailwind, shadcn/ui, zustand, ffmpeg.wasm, opencc-js, docx, and jspdf.

Architecture notes

• /api/transcribe proxies the multipart upload to ${WHISPER_GATEWAY_URL}/v1/audio/transcriptions with advanced=true and diarize=true, streaming NDJSON keepalives so Cloudflare doesn't 524 mid-job.
• /api/report POSTs the transcript to ${OLLAMA_URL}/api/chat with ${OLLAMA_MODEL} (default gemma4:e4b). The prompt explicitly names the target language so the report is written in the right script.
• lib/ffmpeg-client.ts loads @ffmpeg/core from a CDN and converts the user's video to a 16 kHz mono MP3 entirely on the client.
• lib/chinese-convert.ts runs every Chinese segment through OpenCC after transcription, so the user can pick Traditional or Simplified consistently — Whisper itself has no zh-Hant/zh-Hans distinction.

Local development

npm install
npm run dev

Set WHISPER_GATEWAY_URL and OLLAMA_URL to reachable services — see .env.example.

Docker

docker compose up -d --build

Joins the existing infra-net and resolves whisper-gateway:5000 and ollama:11434 inside the network.