What Patchbay Voice is
Patchbay Voice is a voice interface to your codebase, powered by pi, a multi-model AI coding agent. You hold a button and ask about your code, hear the answer, and have the agent save notes for you. It is a developer tool: the app is a client for a small open-source server that you run yourself on macOS or Linux, alongside pi.
Requirements
- A machine running the open-source Patchbay Voice server (MIT licensed) on macOS or Linux.
- pi installed on that machine.
- The app and the server reachable on the same network, or over Tailscale.
Getting started
- Install and start the server on your machine. Setup instructions are in the repository: github.com/synodic-studio/patchbay-voice.
- In the app, open Settings and enter your server’s URL (and an access token if you set one).
- Pick a repository to work in, then hold the microphone button and speak, or switch to the keyboard for precise input.
Common questions
- The app can’t reach the server. Confirm the server is running, that the URL and port in Settings are correct, and that the phone can reach the server on your network. If you set a token on the server, it must match the token in Settings.
- No spoken reply comes back. Check that “Spoken replies” is on in Settings. The app is designed to always return something audible; if speech synthesis fails, you will still hear a short notice.
- Nothing was transcribed. Speak clearly while holding the button and release when done. Very short or silent clips produce no transcript.