Discover the toolingbehind modern intelligence
The operating system for AI infrastructure discovery. Search a live map of MCP servers, AI agents, LLM tools, automation systems, and developer infrastructure.
voice-mcp
Bidirectional voice MCP server for Claude Code — listen (STT) and speak (TTS) on Apple Silicon via mlx-audio
15stars
1forks
Python
Quality signals
StaleNo testsNo CIThin docsNo licenseNew
AI summary
Auto-generated overview of what this skill does and when to reach for it.
Local voice interface for Claude Code on Apple Silicon: listen to speech and have responses spoken back via Voxtral STT and Kokoro TTS.
Use this when
- Debugging code while hands are busy
- Explaining complex concepts aloud to Claude
- Taking voice notes during meetings
- Pair programming with a voice interface
- Testing audio output without leaving the terminal
Don't use when
- Need real-time translation
- Running on non-Apple Silicon hardware
- Requiring ultra-low latency (<100ms) audio
Capabilities
- listen() tool for speech-to-text transcription
- speak() tool for text-to-speech output
- 54 voices across 9 languages via Kokoro TTS
- VAD-based recording with 1.5s silence timeout
- Mic/speaker I/O handled locally on Apple Silicon
- Auto-discovery via .mcp.json for Claude Code
InstallMCPRun via uvx (Python)
Universal installANY TOOL
$ npx skillcity@latest shreyaskarnik/voice-mcpAuto-detects the skill or MCP server in this repo and installs it for Claude Code, Codex, Cursor or Cline.
Requires Node.js 18+ (ships with npx).
Or install manually
uvx voice-mcp
MCP Configuration
Add this to your claude_desktop_config.json
{
"mcpServers": {
"voice-mcp": {
"command": "uvx",
"args": [
"voice-mcp"
]
}
}
}Readme
Loading README…