Changelog

All notable changes to EasySpeak are documented here. This file is the canonical, GitHub-independent record of releases. It is updated once per release. The format loosely follows Keep a Changelog, and the project adheres to Semantic Versioning.

0.4.0 · Packaged and Polished · 2026-06-24

Native Packages & a Docs Site

EasySpeak now ships as native Debian (.deb) and Fedora (.rpm) packages, with the language data split into separate packages so you install only the voices you need. Dictation now actually works — backed by a proper AT-SPI backend that reports failures honestly instead of silently doing nothing — and gains a silent, keyboard hold-to-dictate activation for when you'd rather not speak the wake word. The GNOME tray grows an About dialog and a Help entry, and the extension is renamed simply "EasySpeak".

Under the hood, terminal output moved from print() to structured logging, the codebase adopted Ruff's full "ALL" ruleset with curated exceptions, and a batch of post-0.3.0 review findings were cleaned up; EasySpeak also fails gracefully now when an audio device or gdbus is missing. Rounding it out is a hosted MkDocs and Material documentation site on GitHub Pages — complete with screenshots and Git-LFS-tracked media — while CI is streamlined onto main as the dev branch is phased out.

What's Changed

• feat(packaging): native .deb/.rpm with split language data by @bittner in #77
• Make dictation work: provide the AT-SPI backend and report failures honestly by @bittner in #69
• feat(core): silent (keyboard) hold-to-dictate activation by @bittner in #70
• feat(tray): add About dialog and Help to the tray menu by @bittner in #73
• refactor(extension): rename "EasySpeak Grid" to "EasySpeak" by @bittner in #74
• Switch terminal output from print() to logging by @bittner in #68
• Adopt Ruff "ALL" ruleset with curated exceptions by @bittner in #65
• fix: address post-0.3.0 review findings (browser-launch trap, AT-SPI caret, hotkey shutdown) by @bittner in #72
• Fail gracefully when audio device or gdbus is missing by @bittner in #86
• docs: hosted MkDocs + Material + mkdocstrings site on GitHub Pages by @bittner in #71
• Integrate screenshots in documentation by @bittner in #85
• Serve latest docs at site root, check dead links in README by @bittner in #84
• Track media files with Git-LFS, move images to docs/media by @bittner in #83
• Fix docstring markup (reStructuredText ➜ MarkDown) by @bittner in #81
• Trigger CI workflows on main only, phasing out dev by @bittner in #87
• Fix broken docs screenshots, skip demo video in CI by @bittner in #88

Full Changelog: 0.3.0...0.4.0

0.3.0 · Smoother and More Conversational · 2026-06-13

Tray Icon & Instant Feedback

This release puts EasySpeak in your GNOME tray — a status indicator with voice deactivate — and makes it feel snappier by speaking feedback in parallel with carrying out your command: the spoken reply now starts playing while the action runs, instead of only after it has finished. It also adds native on-screen displays and chimes for volume and brightness via media keys, and a persistent Piper process that loads the voice model once for faster speech. EasySpeak now owns the full GNOME-extension lifecycle — shipped as package data, staged so a failed refresh can't corrupt an install, and time-bounded against hanging system calls. The interaction model grew more conversational too: it keeps listening between commands without repeating the wake word, handles "stop" gracefully instead of quitting, and gives friendlier misunderstanding feedback — backed by new integration/acceptance test tiers and a 99% coverage floor.

What's Changed

• feat(apps): add more GNOME apps, handle default terminal by @bittner in #57
• perf(tts): load the voice model once via a persistent piper process by @bittner in #56
• refactor(core): new speech module, suppress ALSA error output by @bittner in #58
• feat(system): native OSD for volume and brightness via media keys by @bittner in #61
• feat(core): GNOME tray indicator with voice deactivate by @bittner in #59
• test(extension): extract pure JS helpers and unit-test them by @bittner in #60
• Improve voice interaction: friendlier feedback, parallel replies, richer commands, hands-free chaining by @bittner in #63

Full Changelog: 0.2.0...0.3.0

0.2.0 · Broader Reach · 2026-06-07

Runs in More Places

EasySpeak broadens where and how it runs: installation now works on Python 3.13 (with dependencies pinned for 3.10), pyaudio ships as a dependency, and a new Nix flake supports development and running on NixOS. This release also makes the Piper TTS model path configurable, auto-installs the MouseGrid GNOME extension on first run, and adds Whisper transcription-latency benchmarks. Config handling was extracted into its own module and host-environment setup moved into the plugins, tidying the codebase alongside steady dependency and security-workflow upkeep.

What's Changed

• Allow installing on Python 3.13, pin deps for 3.10 by @bittner in #33
• Add Python package version badge to README by @bittner in #34
• Add pyaudio as a package dependency by @bittner in #36
• Update outdated package dependencies (typer-slim ++) by @bittner in #38
• Update outdated package dependencies by @bittner in #39
• Run safety workflow also in PRs by @bittner in #40
• Update outdated dependencies, restrict schedule for check by @bittner in #41
• Update outdated package dependencies by @bittner in #43
• Fix missing apt update for apt install by @bittner in #44
• Update dependencies and actions by @bittner in #45
• Update outdated package dependencies by @bittner in #49
• Add Nix flake for development and running on NixOS by @bittner in #51
• Make Piper TTS model path configurable via env var by @bittner in #52
• Run performance benchmarks for Whisper transcribe latency by @bittner in #53
• Extract config module and move host-env setup into plugins by @bittner in #54
• Fix No plugins directory found error by @gband85 in #47
• Fix mousegrid error caused by duplicate import by @wkarl in #48

New Contributors

• @gband85 made their first contribution in #47
• @wkarl made their first contribution in #48

Full Changelog: 0.1.0...0.2.0

0.1.0 · Foundation · 2026-02-08

Hello, Jarvis

The first release of EasySpeak, published to PyPI as easyspeak-linux: a fully local, Wayland-native voice control for Linux desktops. Say "Hey Jarvis" to drive GNOME hands-free with wake-word activation, a mouse grid, browser control, dictation, and an app launcher — no cloud and no accounts. This release also laid the project's engineering foundation: a clean src layout, a GHA CI pipeline for linting, typing, tests, and an automated PyPI publish, and a first suite of unit tests covering the core engine, CLI, and browser plugin.

What's Changed

• chore(gitignore): change gitignore file to .gitignore and remove pre-compiled python assets by @tulilirockz in #11
• Add a Justfile, draft packaging setup, contributing docs by @bittner in #14
• Organize Python modules with src layout, add first tests by @bittner in #15
• Add CI pipeline for linting and tests by @bittner in #17
• Allow all CI jobs to finish, reformat codebase, fix linting by @bittner in #18
• Fix type check config, make tests run and pass by @bittner in #19
• Update dependencies (uv.lock) by @bittner in #20
• Add unit tests for core module (mostly EasySpeak class) by @bittner in #21
• Run software safety related jobs only once a day by @bittner in #23
• Refactor application entrypoint (CLI), add tests by @bittner in #22
• Add tests for browser plugin by @bittner in #25
• Update outdated package dependencies by @bittner in #26
• Add tests for dictation plugin by @bittner in #27
• Add tests for apps plugin by @bittner in #28
• Add remaining tests for 100% coverage by @bittner in #29
• Update outdated package dependencies by @bittner in #30
• Build Python package, configure automatic release by @bittner in #32

New Contributors

• @tulilirockz made their first contribution in #11
• @bittner made their first contribution in #14

Full Changelog: 0.1.0