OpenBrief: Open-Source Local Video Summarizer
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- OpenBrief is a Tauri v2 desktop app (pnpm/Turborepo monorepo) that imports local media or video URLs, downloads via bundled tools, transcribes audio, generates grounded markdown summaries, and supports chat against the transcript.
- OpenBrief runs entirely on the user's machine under the AGPL v3.0 license, using local speech-to-text (whisper.cpp, transcribe-rs) and on-device TTS rather than cloud APIs.
- The project sits in a monorepo with a Tauri desktop app, a Next.js web app, a TanStack Start shell, an Expo React Native shell, and shared packages for API routing, auth, database, UI, and validators.
- Media processing is built on yt-dlp for downloads, whisper.cpp and transcribe-rs for on-device speech-to-text, FluidAudio for Apple-platform audio AI, and Qwen3-ASR, Qwen3-TTS, and Supertonic for model support.
- The roadmap lists Parakeet ASR, Qwen3-ASR and Qwen3-ForcedAligner, Supertonic 3 TTS, local LLM support including Gemma 4, video frame and clip semantic search, voice cloning, PDF and HTML document import, and flashcard-style exports.
- Setup requires Node.js ^22.21.0, pnpm 11.0.9, Rust/Cargo, and Tauri v2 platform prerequisites, with
pnpm dev:taurilaunching the desktop window andpnpm dev:nextrunning the web app on localhost:3000.
Why it matters: OpenBrief packages a full local media-to-summary pipeline — download, transcribe, summarize, chat, and export — into one free AGPL-licensed desktop app, giving privacy-conscious users a self-hosted alternative to cloud summarization tools, though the roadmap (local LLMs, Parakeet ASR, video semantic search) shows most of the differentiating features are still in progress.
Ask SkimNews
