VoiceMode MCP
An MCP server for voice conversations with Claude Code and other MCP agents: speech recognition and voice synthesis, local or cloud
- 1,454 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Accesses the microphone and plays audio on the device
- In cloud mode it sends audio to an external recognition and synthesis service
- Installs and runs local speech services
Install
- Works in
Manual install
claude plugin marketplace add mbailey/voicemode
claude plugin install voicemode@voicemodeAfter install run /voicemode:install for dependencies and local services, then /voicemode:converse.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Accesses the microphone and plays audio on the device
- In cloud mode it sends audio to an external recognition and synthesis service
- Installs and runs local speech services
This is third-party code. Review the repository files before installing.
What it does
VoiceMode gives the agent a converse tool: you speak into the microphone and the agent answers by voice. Speech recognition and synthesis can run fully local via Whisper and Kokoro, or use cloud services behind an OpenAI-compatible API, and it switches between them transparently. It detects silence to stop recording when you pause and focuses on low latency so the exchange feels like a real conversation. The server installs as a Claude Code plugin or as an MCP server through uvx, and runs on Linux, macOS and Windows including WSL. The package is published on PyPI as voice-mode.
Who it is for. For developers who prefer to talk to the agent when their hands or eyes are busy.
Good fit when
- You want to talk to Claude Code by voice instead of typing
- You need to work with the agent away from the screen: on the move, while cooking, when your eyes are tired
- You need local voice input and output without sending audio to the cloud
Not a fit when
- There is no microphone or speakers, or the environment has no audio access
- You only need text input and do not need voice
Example request
Let's switch to voice mode and talk through how to rewrite this moduleMCP
- Transport
- stdio
- Authentication
- not required
Environment variables
OPENAI_API_KEYsecret- OpenAI key for cloud speech recognition and synthesis, optional; the local mode works without it.
Limitations
You need a working microphone and speakers. Local speech services need ffmpeg, portaudio and similar dependencies, plus disk space and compute for the models. The cloud path through OpenAI needs a key and is not reachable from Russia without a VPN, while the local mode works without one. On WSL2 microphone access needs the pulseaudio packages.
How to disable. Remove the plugin via /plugin or detach the MCP server with claude mcp remove voicemode. The local speech services can be stopped and removed separately.
FAQ
Can it run offline without an OpenAI key?
Yes. With local Whisper and Kokoro the recognition and synthesis run on your machine. The OpenAI key is only for the cloud path and acts as a fallback.
Which agents does it work with?
The main path is the Claude Code plugin, but the server speaks the MCP protocol, so it also connects to other MCP-compatible agents.
Discussion
No questions or reviews yet.
Related
MCP serversA realtime voice layer on top of Claude Code, Codex and OpenCode that keeps talking and listening while background tasks run in parallel
Local voice layer for Claude Code and Codex on macOS: when the agent finishes or fails, it says in a sentence or two what happened
Local AI commands for voice and text: clipboard dictation, text fixes, speech, an assistant and parallel agents in worktrees
A local MCP server that lets the agent ask you questions in the console and notify you when a task is done
Local dictation for macOS with Claude Code and Codex plugins: speech becomes prompt text that recognizes your files.
Minutes
minutes
A local-first MCP server for meeting memory: on-device recording and transcription into Markdown, searchable from Claude Code, Codex, Cursor and any MCP client
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| VoiceMode (this card) | yes | not required | 1,454 | |
| ElevenLabs Skills | yes | 559 | ten skills from ElevenLabs | |
| yovoice | no | 2,173 | local speech synthesis for agents | |
| JARVIS setup | no | 978 | A skill helps set up voice JARVIS for Claude Code on a Mac and check microphone access, speech synthesis and permissions |
README in short
The README presents VoiceMode as a way to hold voice conversations with Claude Code and other MCP agents. The quick start offers two paths: the Claude Code plugin via its marketplace, or installing the Python package and wiring the MCP server through uvx. Features include natural dialogue, offline use with local speech services, low latency, silence detection and a choice between local and cloud modes. Separate sections cover system dependencies per platform, permission setup, installing Whisper and Kokoro, and common audio problems. It is MIT licensed and the PyPI package is named voice-mode.