A fork of OpenWhispr adding native support for COSMIC on Pop!_OS 24.04 and a hands-free voice-agent workflow.
This is a community fork, not the official OpenWhispr project. Built and tested on Pop!_OS 24.04 specifically — other COSMIC-based distros or versions may work but aren't verified. For macOS, Windows, or non-COSMIC Linux desktops, use upstream: OpenWhispr/openwhispr.
No support is provided. This is shared as-is for anyone who finds it useful. Issues and PRs may not get a response — read the code, fork it, and adapt it yourself if something breaks. For the officially supported app, use upstream.
Upstream OpenWhispr already has native global-hotkey backends for GNOME, Hyprland, and KDE Wayland, but nothing for COSMIC — Electron's built-in globalShortcut API doesn't fire for native Wayland clients there, so the hotkeys silently didn't work on Pop!_OS 24.04. This fork adds:
- Native COSMIC global hotkeys — dictation, AI agent, meeting, and voice-agent hotkeys all register directly through COSMIC's own shortcuts config, at parity with the existing GNOME/Hyprland/KDE backends.
launch_applicationagent tool — the AI agent can open installed applications by voice ("open Firefox"), matched against your installed.desktopentries.- Hands-free wake-word routing — saying your agent's wake word during dictation (or using the dedicated Voice Agent hotkey) opens the tool-capable chat overlay instead of just rewriting text in place, so voice commands can actually take action, not just clean up transcripts.
- Vulkan GPU context-size fix — local LLM inference on Vulkan-accelerated GPUs no longer silently falls back to CPU due to an unbounded-context VRAM overrun on launch.
Everything else — transcription engines, meeting detection, notes, cloud/local model providers — is unchanged from upstream. See CUSTOM_WORK.md for the full technical writeup of each change.
The open-source and free alternative to WisprFlow and Granola.
Privacy-first voice-to-text dictation with AI agents, meeting transcription, and notes. Cross-platform for macOS, Windows, and Linux.
Website · Docs · Download · API · Changelog
OpenWhispr turns your voice into text, notes, and actions from your desktop. Press a hotkey, speak, and your words appear at your cursor. Choose between fully private offline transcription with local speech-to-text engines like Whisper and NVIDIA Parakeet — where your audio never leaves your device — or cloud processing for speed. No data collection, no telemetry, fully open source.
| Platform | Download |
|---|---|
| macOS (Apple Silicon) | .dmg |
| macOS (Intel) | .dmg |
| Windows | .exe |
| Linux | .AppImage / .deb / .rpm / .tar.gz |
- Voice dictation — global hotkey to dictate into any app with automatic pasting
- AI agent — talk to GPT-5, Claude, Gemini, Groq, Tinfoil, or local models with a named voice assistant
- Voice agent hotkey — dedicated hotkey that sends your dictation straight to your AI agent as a command, no wake word needed and no cleanup pass
- Meeting transcription — auto-detect Zoom, Teams, and FaceTime calls with live speaker diarization, voice fingerprinting, and Google Calendar integration
- Local speaker diarization — on-device speaker labelling with voice fingerprint recognition across meetings, no cloud required
- Notes — create, organize, and search notes with folders, semantic search, cloud sync, and AI actions
- Local or cloud — your choice — all core features (transcription, AI reasoning, speaker diarization, semantic search) work with local models or cloud providers
- Public API & MCP — manage notes and 8000 transcriptions programmatically or connect your AI assistant via the MCP server
git clone https://github.com/Fozzie-VR/openwhispr.git
cd openwhispr
npm install
npm run devRequires Node.js 24+. See the full documentation for setup guides, platform-specific instructions, and build details.
Visit docs.openwhispr.com for:
- Getting started
- Platform guides (macOS, Windows, Linux)
- API reference
- MCP server setup
- Troubleshooting
React 19, TypeScript, Tailwind CSS v4, Electron 41, better-sqlite3, whisper.cpp, sherpa-onnx, shadcn/ui
Neon is the serverless Postgres platform powering OpenWhispr Cloud.
MIT — free for personal and commercial use.
- OpenAI Whisper — speech recognition model powering local and cloud transcription
- whisper.cpp — high-performance C++ implementation for local processing
- NVIDIA Parakeet — fast multilingual ASR model
- sherpa-onnx — cross-platform ONNX runtime for Parakeet inference
- Hugging Face — model hub hosting Whisper, Parakeet, and embedding model weights
- llama.cpp — local LLM inference for AI text processing
- Electron — cross-platform desktop framework
- React — UI component library
- shadcn/ui — accessible components built on Radix primitives
- Neon — serverless Postgres powering OpenWhispr Cloud