# OpenBrief

> A local-first, open-source desktop app that turns long videos and audio into transcripts, grounded summaries, and exportable notes—no account required.

OpenBrief is an open-source desktop application built with Tauri v2 that converts long videos, audio files, and recordings into structured briefs. It runs entirely on your machine—no sign-up, no cloud dependency—and is licensed under the GNU Affero General Public License v3.0. The project is maintained by GitHub user tantara and reached v0.4.0 in May 2026.

## What It Is

OpenBrief is a local-first briefing workspace in the category of AI-powered media summarization tools. Its core job is to take a video or audio source—whether a local file or a supported web link—and produce a transcript, a grounded summary, and an interactive chat interface, all within a single desktop window. The output can be exported to Markdown for use anywhere. The app is designed for research calls, lectures, product demos, interviews, and screen recordings.

## How the Local-First Architecture Works

Everything OpenBrief generates stays on the user's device. The app bundles helper sidecars for media downloading (via yt-dlp) and on-device speech-to-text (via whisper.cpp and transcribe-rs). When captions already exist on a source, OpenBrief uses them directly and only runs transcription when necessary. LLM calls for summarization and chat are routed through user-supplied API keys to providers such as OpenAI GPT, Anthropic Claude, Google Gemini, and OpenRouter DeepSeek—no data passes through OpenBrief's own servers.

## Model Support and AI Capabilities

The README documents the following model support matrix:

- **Speech to text**: Whisper, Parakeet, Qwen3-ASR
- **Text to speech**: Supertonic 3, Qwen3-TTS
- **Large language models**: OpenAI GPT, Anthropic Claude, Google Gemini, OpenRouter DeepSeek
- **Local LLMs**: Gemma 4 is listed on the roadmap but not yet supported

Grounded summaries are generated as blog-style Markdown briefs with timestamped takeaways, and chat answers are tied directly back to the transcript rather than generated from general knowledge.

## Repository Layout and Tech Stack

OpenBrief is a pnpm/Turborepo monorepo. The primary deliverable is a Tauri v2 desktop app with a React renderer and a Rust backend. The repository also contains shells for a Next.js web app, a TanStack Start app, a React Native (Expo) app, and worker entry points—indicating a planned multi-platform expansion. The desktop app requires Node.js 22.21.0+, pnpm 11.0.9, Rust/Cargo, and Tauri v2 platform prerequisites.

## Update: OpenBrief v0.4.0

The latest release is v0.4.0, published on May 25, 2026. Recent completed roadmap items include improved audio file support for transcription, summaries, playback, and exports; Parakeet ASR support; Qwen3-ASR and Qwen3-ForcedAligner support; and Supertonic 3 TTS support. Upcoming roadmap items include support for local LLMs (Gemma 4), PDF and HTML document sources, video frame/clip embeddings for semantic search, voice cloning, web and mobile sharing, and additional artifact formats such as flashcards.

## Features
- Import local video/audio files or paste a web video link
- Extract captions or run on-device speech-to-text transcription
- Generate grounded Markdown summaries with timestamped takeaways
- Chat with media content grounded in the transcript
- Text-to-speech playback of summaries
- Export notes to Markdown
- Searchable library of all imported sources
- No account or sign-up required
- All files and generated notes stay on-device
- Bring-your-own API keys for LLM providers

## Integrations
OpenAI GPT, Anthropic Claude, Google Gemini, OpenRouter DeepSeek, yt-dlp, whisper.cpp, Qwen3-ASR, Qwen3-TTS, Supertonic 3, Parakeet ASR

## Platforms
WINDOWS, MACOS, LINUX, WEB, API

## Pricing
Open Source

## Version
v0.4.0

## Links
- Website: https://openbrief-phi.vercel.app
- Repository: https://github.com/tantara/openbrief
- EveryDev.ai: https://www.everydev.ai/tools/openbrief