A fast, accurate, and private native speech-to-text tool for Linux, offering system-wide dictation with local or cloud backends.
hyprwhspr is a native speech-to-text tool for Linux that provides fast, accurate, and private system-wide dictation. It solves the problem of needing a reliable, locally-run dictation solution by supporting both local AI models and cloud APIs, enabling users to transcribe speech into text instantly across any application.
Linux users, particularly those on Wayland-based desktops like GNOME, KDE Plasma, Sway, or Hyprland, who need efficient and private dictation for productivity, accessibility, or hands-free input.
Developers choose hyprwhspr for its combination of local performance, privacy, and deep system integration, offering a seamless dictation experience without relying on external services, along with support for the latest AI models and customizable features.
Native speech-to-text for Linux - Fast, accurate and private system-wide dictation
Uses in-memory models for instant transcription without external servers, ensuring data never leaves your machine, as highlighted by the 'local, very fast defaults' in the README.
Supports Cohere Transcribe, Parakeet TDT V3, Whisper, REST APIs, and WebSockets, allowing users to choose between local and cloud-based transcription for varied needs.
Efficient GPU memory usage and CPU-optimized via onnx-asr provide great speed even on systems without dedicated graphics, per the README's emphasis on performance.
Auto-pastes transcribed text into any active buffer and integrates with Waybar and custom hotkeys, enabling seamless desktop workflow without manual copying.
Only works on Wayland sessions, excluding many Linux users on X11-based desktops, which limits its usability across common setups like traditional GNOME or KDE.
Requires interactive setup, systemd service configuration, and permission management, which can be intimidating for less technical users despite the guided process.
Non-Arch Linux support is explicitly noted as new and may have unresolved issues, requiring user reporting and patience for stability.
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.