MCP MCP server · dev.voicemode
dev.voicemode/voicemode
Natural voice conversations for AI assistants — STT/TTS via MCP
- host stars
- 1,368
- 30-day movement
- +83/day
- Related entries
- 2
- Connections
- 0
pypilocalstdio
VoiceMode is an MCP server that adds natural voice conversations to AI assistants like Claude Code, handling speech-to-text and text-to-speech over stdio. It works out of the box with a microphone and speakers, and can optionally use local Whisper STT and Kokoro TTS services for offline or privacy-focused use.
You want to speak to and hear responses from your MCP-capable coding agent instead of typing.
Use it to
- Talk to Claude Code hands-free
- Run voice conversations fully offline with local Whisper and Kokoro
- Set up voice input on Linux, macOS, Windows or WSL
- Configure Claude Code permissions to skip voice prompts
- Debug audio issues by saving captured audio
For Developers using Claude Code or other MCP agents
- Runs
- installed — on your machine
- Transport
- stdio
- Packaged as
- pypi:voice-mode
- Install
- uvx voice-mode
- Needs
- 7 environment variables, including secrets
- Version
- 8.12.0
- Last release
- 2026-07-21
topicsvoicespeech-to-texttext-to-speechmcp-serverclaude-codeoffline