🛸 Atlas v0.2.3 — Voice Control, Settings Redesign & Smart Defaults
Talk to Atlas hands-free and enjoy a completely rebuilt settings experience.
What's new
- 🎙️ Local Speech-to-Text — offline voice recognition via Vosk, no cloud API required. Just say the persona's name and start talking
- 👂 Wake word activation — Atlas listens for the active persona's name and starts capturing your command automatically
- 🏝️ Listening Island — a floating UI shows your live transcript as you speak, so you know exactly what Atlas hears
- 🖥️ Full-page Settings — the settings panel has been rebuilt from scratch with sidebar navigation, auto-save, and a cleaner layout
- 🧠 Intelligence tab — LLM and generation parameters are now grouped under a single "Intelligence" tab
- 🔊 Voice tab — TTS and STT settings live together in one unified "Voice" tab
- 😀 Emoji personas — pick an emoji avatar for each persona with the new built-in emoji picker
- 🔄 Reset to defaults — every settings section now has a one-click reset button
- 🖱️ Multi-monitor screenshots — vision actions now target the correct display on multi-monitor setups
- 🔇 TTS interruption fix — voice playback stops immediately when you interrupt the agent
Full changelog
See CHANGELOG for a detailed list of all added, changed, and removed items.
⚠️ Windows only. macOS & Linux support is planned.