Equips AI agents with voice input and output capabilities, acting as a bridge for natural language interaction.
Sponsored
Agent Vox is a lightweight, local voice service designed to give AI agents, such as Claude Code, Codex, and Gemini CLI, the ability to understand spoken language and respond verbally. It serves as a voice bridge, providing 'ears' for Automatic Speech Recognition (ASR) and 'mouth' for Text-to-Speech (TTS), without including the Large Language Model (LLM) itself. The tool supports multiple integration methods including MCP for native agent tool calls, WebSocket for custom application integration, and a CLI for single-use scripting.
Key Features
01Utilizes free Microsoft Edge TTS for natural-sounding voice synthesis with various Chinese voices.
020 GitHub stars
03Integrate voice (ASR & TTS) with AI agents like Claude Code, Codex, and Gemini CLI.
04Provides local, privacy-focused ASR using whisper-cli for robust multilingual support.
05Built with Bun for fast and lightweight execution, prioritizing macOS Apple Silicon.
06Offers three primary access methods: MCP (recommended), WebSocket, and CLI.
Use Cases
01Integrating real-time voice capabilities into front-end applications or custom AI agent interfaces via WebSocket.
02Enabling native voice interaction for AI agents that support MCP tool calls.
03Adding speech-to-text or text-to-speech functionality to scripts or agents that interact via command-line.
