brainstate

pavlenkosa/wayvoice · source

Snapshot 0 · 2026-11-04T12:47:28Z · Python · AGPL-3.0

Content fingerprintef41e818ac85650f782f6954016180d151ee7142d448da28028c28e19e6145bd

GitHub description at capture

Offline voice typing for Linux/Wayland. Press a hotkey, speak, and text appears in the active application. Local Whisper • GTK/libadwaita • No cloud required.

README at capture

<p align="center">
  <img src="data/icons/hicolor/scalable/apps/io.github.stepan.WayVoice.svg" width="96" alt="WayVoice logo">
</p>

# WayVoice

**Private, offline voice typing for Linux.** Press a shortcut. Speak. Keep typing.

[![CI](https://github.com/pavlenkosa/wayvoice/actions/workflows/ci.yml/badge.svg)](https://github.com/pavlenkosa/wayvoice/actions/workflows/ci.yml)
[![License: AGPL-3.0-or-later](https://img.shields.io/badge/license-AGPL--3.0--or--later-blue)](LICENSE)
![Status: Beta](https://img.shields.io/badge/status-beta-yellow)
[![XERJ Corpus Hub contributor](https://hub.xerj.org/contributors/pavlenkosa/badge.svg)](https://hub.xerj.org/contributors/pavlenkosa)

[Download](https://github.com/pavlenkosa/wayvoice/releases) · [Getting started](docs/getting-started.md) · [Changelog](CHANGELOG.md) · [Get help](SUPPORT.md)

![WayVoice English Home interface with recording controls and preparation checks](data/screenshots/home-en.png)

Real English interface shown with an isolated demo state, without personal dictation. [Settings](data/screenshots/settings-en.png) · [First launch](data/screenshots/setup-en.png)

## Install

Open [GitHub Releases](https://github.com/pavlenkosa/wayvoice/releases), choose a release and download a package listed in its assets. Verify it against that release's `SHA256SUMS`. Beta releases are marked **Pre-release** and are installed manually; the native updater follows stable releases only.

```bash
# Debian / Ubuntu
sudo apt install ./wayvoice_*_amd64.deb

# Fedora 44
sudo dnf install ./wayvoice-*-1.fc44.x86_64.rpm
```

Open **WayVoice**, explicitly prepare a model, and configure your shortcut. Initial runtime/model preparation needs a network connection; recognition with prepared local models works offline. Missing models are never downloaded by pressing the dictation shortcut.

Version 0.8.0 replaces the experimental Flatpak build with an x86_64 AppImage containing Python, GTK/libadwaita and recognition runtimes. It targets Linux with glibc 2.39 or newer. See [portable installation and manual updates](docs/installation.md#appimage); live desktop acceptance remains open.

## Features

- Local Faster-Whisper and whisper.cpp recognition; optional warm recognizer.
- Wayland clipboard delivery and optional automatic paste. If insertion is unavailable, text stays in the clipboard.
- Explicit Faster-Whisper model preparation with progress, cancellation and storage controls; quick prepared model/language selection on Home.
- Microphone selection, optional recording sounds, personal text replacements and a compact status indicator.
- Experimental local CPU Parakeet v3 with explicit preparation and automatic detection across 25 model languages. See [engine details and limitations](docs/engines-and-models.md).
- Ten offline interface languages: English, Russian, Spanish, Portuguese, French, German, Simplified Chinese, Japanese, Arabic and Hindi.
- Automatic speech-language detection or fixed language with compatible models. See [all 100 recognition language choices and model limitations](docs/engines-and-models.md#supported-languages).

## Desktop support

**Beta:** automated tests do not establish complete desktop support.

| Environment | Current scope |
| --- | --- |
| GNOME / Wayland | Primary tested desktop |
| KDE Plasma / Wayland | Shortcut integration implemented; complete installed dictation/paste acceptance pending |
| wlroots | Community testing needed |
| AppImage | x86_64, glibc ≥ 2.39; portable desktop acceptance pending |

Automatic paste depends on permissions and optional `ydotool`; manual clipboard paste remains available. GNOME shortcuts toggle recording; portal hold-to-talk currently targets KDE. [Desktop compatibility](docs/desktop-support.md) · [Shortcuts and paste](docs/shortcuts-and-paste.md)

## Documentation

[Installation and updates](docs/installation.md) · [Getting started and CLI](docs/getting-started.md) · [Engines, models and languages](docs/engines-and-models.md) · [Troubleshooting and privacy](docs/troubleshooting.md)

For contributors: [Contributing](CONTRIBUTING.md) · [Development](docs/development.md) · [Architecture](docs/architecture.md) · [Release process](docs/releasing.md) · [Security](SECURITY.md)

Bug reports, translations and desktop testing are welcome. Review diagnostics and remove private dictation before sharing them.

## License

[GNU Affero General Public License v3.0 or later](LICENSE). Bundled helper licensing is listed in [third_party/ydotool](third_party/ydotool/README.wayvoice.md).