Banshee

•
19
•

Empowers AI coding agents with voice communication, enabling spoken interactions for decisions, questions, and desktop control, all hands-free.

Banshee transforms the interaction with AI coding agents by giving them a voice, allowing them to articulate decisions and questions aloud. This innovative tool facilitates a completely hands-free workflow, where users respond through speech, creating a seamless conversational development environment. It prioritizes privacy by operating locally with Whisper for speech-to-text and a neural voice for text-to-speech, ensuring user data remains on their device. Beyond its role in enhancing AI agent collaboration, Banshee also serves as a versatile dictation utility, typing spoken words directly into any focused application, and empowers users to manage and modify their desktop environment using simple voice commands.

Key Features

01Hands-free desktop control through voice commands.
02General dictation tool for typing spoken words into any focused application.
0313 GitHub stars
04Local-first operation with Whisper (speech-to-text) and a local neural voice (text-to-speech) by default.
05Integrates with various coding agents (e.g., Claude Code, Antigravity, OpenCode, Pi, Codex, Cursor) via MCP server shim.
06Voice interface for AI coding agents: agents speak decisions/questions, users answer by talking.

Use Cases

01Interacting with AI coding agents for development tasks using natural spoken language.
02Dictating text into any application quickly without needing to type.
03Controlling and modifying desktop environments hands-free for enhanced productivity.