# Lip Service

> Lip Service is an open-source MCP server for audio processing. Offline text-to-speech and singing synthesis that runs on your own CPU, for Python and CLI. It runs locally on your machine and is started by your AI client.

- **Kind:** MCP server
- **Pricing:** Open source
- **Platforms:** API, CLI, Linux, macOS, Web, Self-hosted
- **Website:** https://lipservice.dev
- **MCP Registry name:** dev.lipservice/lipservice
- **Categories:** [audio processing](https://ai-powered.com/categories/audio-processing)
- **Listing updated:** 2026-11-19

## About

The README describes Lip Service as a speech and singing engine that runs entirely on local hardware, with sixteen English voices that can either play audio aloud or write it to a WAV file. Input can be plain text for speech or a simple note-and-lyric script for singing, and the site notes that MIDI files can be imported and exported with their lyrics.

The project ships in several forms: a Python package installable with pip, a command-line tool with say and sing subcommands, a browser build running through WebAssembly, and a one-line embed script that adds a Listen button to a web page. The site says the engine has no third-party dependencies and that the first run downloads about 56 MB of models, which are then kept locally.

It suits developers who want speech or song output inside scripts, local tools or MCP-connected assistants without sending text to a cloud service. Licensing is AGPL for free use, with a separate commercial license offered for closed-source or hosted products.

## Features

- **Sixteen local voices:** Offers sixteen English voices that sing or speak on the user's own machine.
- **Singing from note scripts:** Turns a text score with pitches, durations and lyrics into sung audio, with multi-singer mixing.
- **Python and CLI access:** Exposes speech and singing through a Python module and a lipservice command-line tool.
- **WAV and MIDI handling:** Writes results to WAV files and imports or exports MIDI files with lyrics.
- **Browser embed option:** Runs in a browser via WebAssembly and can add a read-aloud button with one script tag.
- **MCP output folder setting:** The MCP server writes WAV files to a folder set by LIPSERVICE_OUT; calls with no device or path are refused when it is unset.

## Good for

- Generating narration or spoken alerts from a script without a cloud speech API
- Producing short sung clips from a lyric and note file for prototypes or demos
- Adding read-aloud support to a web page with a single embed script
- Letting an MCP-connected AI assistant save spoken or sung WAV files locally

## Install

### Claude Code (Terminal)

```sh
claude mcp add lipservice -- uvx lipservice
```

### Claude Desktop (claude_desktop_config.json)

```json
{
  "mcpServers": {
    "lipservice": {
      "command": "uvx",
      "args": [
        "lipservice"
      ]
    }
  }
}
```

### Cursor (~/.cursor/mcp.json)

```json
{
  "mcpServers": {
    "lipservice": {
      "command": "uvx",
      "args": [
        "lipservice"
      ]
    }
  }
}
```

### VS Code (.vscode/mcp.json)

```json
{
  "servers": {
    "lipservice": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "lipservice"
      ]
    }
  }
}
```

### Windsurf (~/.codeium/windsurf/mcp_config.json)

```json
{
  "mcpServers": {
    "lipservice": {
      "command": "uvx",
      "args": [
        "lipservice"
      ]
    }
  }
}
```

### Other clients (your client's MCP settings (most use this shape))

```json
{
  "mcpServers": {
    "lipservice": {
      "command": "uvx",
      "args": [
        "lipservice"
      ]
    }
  }
}
```

## FAQ

### Does Lip Service need an internet connection to run?

The site says the first call fetches the roughly 56 MB of models and keeps them, so later use runs on the local CPU.

### Which platforms are supported natively?

The site says native builds run on macOS and Linux, and a browser version runs through WebAssembly.

### What license does Lip Service use?

It is free under the GNU AGPL. The site offers a commercial license for closed products or services.

### How is the MCP server configured?

The registry entry lists the LIPSERVICE_OUT environment variable, which sets the folder for WAV files when a call names no device or full path.

### Is Lip Service free?

Yes. Lip Service is open source, so you can use and self-host it for free.

### How do I add Lip Service to Claude Code?

Run this command in your terminal: claude mcp add lipservice -- uvx lipservice

### What are the best alternatives to Lip Service?

The closest alternatives to Lip Service are Audio Studio, Saidwire Speech-to-Text, Voicely, Elevenlabs and Live Translate MCP.

## Where this comes from

- Description: written from the project's website and its MCP Registry entry.
- Install steps: from the official MCP Registry entry.

## Alternatives

- [Audio Studio](https://ai-powered.com/mcp/audio-studio): Audio Studio is an open-source MCP server for design and media.
- [Saidwire Speech-to-Text](https://ai-powered.com/mcp/saidwire-speech-to-text): Saidwire Speech-to-Text is a freemium MCP server for design and media.
- [Voicely](https://ai-powered.com/mcp/voicely): Voicely is an open-source MCP server for design and media.
- [Elevenlabs](https://ai-powered.com/mcp/elevenlabs-4): Elevenlabs is an open-source MCP server for design and media.
- [Live Translate MCP](https://ai-powered.com/mcp/live-translate-mcp-2): Live Translate MCP is an open-source MCP server for audio processing.
- [GUI Hub Audio](https://ai-powered.com/mcp/gui-hub-audio): GUI Hub Audio is an MCP server for design and media.
- [Whisper Transcribe](https://ai-powered.com/mcp/whisper-transcribe): Whisper Transcribe is an open-source MCP server for audio processing.
- [MCP Listen](https://ai-powered.com/mcp/mcp-listen-2): MCP Listen is an open-source MCP server for audio processing.

---
Source: https://ai-powered.com/mcp/lip-service
