Tools / repository

voice-os

voice-os is a hackable macOS voice-control starter that uses OpenAI gpt-realtime-2 for speech-to-speech with tool calling and agent-desktop to drive apps via the accessibility tree. Shipped tools open apps, play Spotify, run terminal commands, read the screen, control Premiere, and start OBS recordings.

URL: https://github.com/per-simmons/voice-os
Collection: Design & Content
License: MIT
Open source: Yes

macos voice-control speech-to-speech openai-realtime accessibility automation agent-desktop python github open-source design media