mcp-registry/voicemode

    ==================
      
       /// MCP ///
      /// VOI ///
        
    ==================
        
    [server:online]
    [protocol:ready]

voicemode

by mbailey

Natural voice conversation MCP server enabling real-time STT/TTS talking to Claude Code and other AI coding assistants.

149
16
Open Source
GitHubWebsite

01

converse

Have a voice conversation—speak and optionally listen to responses

02

listen_for_speech

Listen for speech and convert it to text

03

check_room_status

Check the status of a LiveKit room and its participants

04

check_audio_devices

List available audio input and output devices

05

start_kokoro

Start the local Kokoro text-to-speech service

06

stop_kokoro

Stop the Kokoro text-to-speech service

07

kokoro_status

Report the current status of the Kokoro text-to-speech service

08

install_whisper_cpp

Install whisper.cpp for running local speech-to-text

09

install_kokoro_fastapi

Install kokoro-fastapi for running local text-to-speech

Installation

1. Prerequisites
• Python ≥ 3.10
• Git, Make (optional) and ffmpeg installed and discoverable in PATH (ffmpeg is required for Whisper and most TTS back-ends).
• A working microphone if you plan to use live voice capture.
• Access tokens:
- OPENAI_API_KEY ─ required for Whisper or GPT-based agents.
- LIVEKIT_URL and LIVEKIT_API_KEY/LIVEKIT_API_SECRET ─ if you want remote audio transport.
2. Clone the repository
git clone https://github.com/mbailey/voicemode.git cd voicemode
3. Create and activate a virtual environment (recommended)
python -m venv .venv source .venv/bin/activate # Windows: .venv\Scripts\activate
4. Install Python dependencies
pip install -U pip pip install -r requirements.txt # optional development extras pip install -r requirements-dev.txt
5. (Optional) Build native Whisper kernels for faster inference
make whisper # or follow the README notes if Make isn’t available.
6. Configuration
• Copy the example environment file and edit values:
• Alternatively, export variables directly in your shell:
cp .env.example .env # open .env and set OPENAI_API_KEY and other service credentials export OPENAI_API_KEY="sk-..." export LIVEKIT_URL="wss://..."
7. Run the MCP server locally
python -m voicemode.server --host 0.0.0.0 --port 8080 # visit http://localhost:8080 or connect with an MCP client.
8. Docker (optional)
docker build -t voicemode . docker run -p 8080:8080 --env-file .env voicemode
9. Updating
git pull && pip install -r requirements.txt --upgrade

Documentation

License: Unknown – no license file detected in repository metadata
Updated 7/30/2025