# Macwhisper MCP Server

> Macwhisper MCP Server is an open-source MCP server for design and media. Transcribe audio on your Mac with MacWhisper from Claude Desktop, fully local. It runs locally on your machine and is started by your AI client.

- **Kind:** MCP server
- **Pricing:** Open source
- **Platforms:** macOS, CLI, self-hosted
- **Repository:** https://github.com/docdyhr/macwhisper-mcp-server
- **MCP Registry name:** io.github.docdyhr/macwhisper-mcp-server
- **Categories:** [design and media](https://ai-powered.com/categories/design-and-media), [files and documents](https://ai-powered.com/categories/files-and-documents)
- **Listing updated:** 2026-11-18

## About

Audio never leaves the machine. The server calls the MacWhisper command-line tool, so Claude Desktop sends the file path to a local process and gets a transcript back for summaries or action items. It needs macOS, a licensed MacWhisper with its CLI installed, Python 3.13 and Claude Desktop, and it reads only from directories named in MACWHISPER_ALLOWED_PATHS.

Seven tools are listed. transcribe_audio takes a path and optional model, language, persist and engine settings; list_models, list_allowed_paths and cancel_transcription round out control; start_watch, stop_watch and get_watch_results watch a folder and auto-transcribe new files into a done folder. Language defaults can be mapped per directory, and a whisper.cpp engine is available as an alternative when a model directory is configured. Supported formats are m4a, mp3, mp4, mov, wav, aiff and flac.

Files dropped into the Claude chat window are not reachable, since they live in Claude's container, so audio must sit in an allowed folder on the Mac.

## Features

- **Local transcription:** Runs MacWhisper on your Mac with no cloud service.
- **Folder allowlist:** Reads audio only from directories you list.
- **Watch folders:** Auto-transcribes new files into a done folder.
- **Language defaults:** Map folders to languages or use auto-detect.
- **whisper.cpp option:** Alternate engine when GGML models are configured.
- **Model listing:** Shows installed MacWhisper models and the active one.

## Good for

- Transcribe a voice memo and pull out action items
- Summarise a recorded meeting in Claude Desktop
- Batch-transcribe a folder of interviews
- Choose a language per recordings folder

## Install

### Claude Code (Terminal)

```sh
claude mcp add macwhisper-mcp-server --env MACWHISPER_ALLOWED_PATHS='<MACWHISPER_ALLOWED_PATHS>' -- uvx macwhisper-mcp-server
```

### Claude Desktop (claude_desktop_config.json)

```json
{
  "mcpServers": {
    "macwhisper-mcp-server": {
      "command": "uvx",
      "args": [
        "macwhisper-mcp-server"
      ],
      "env": {
        "MACWHISPER_ALLOWED_PATHS": "<MACWHISPER_ALLOWED_PATHS>"
      }
    }
  }
}
```

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

```json
{
  "mcpServers": {
    "macwhisper-mcp-server": {
      "command": "uvx",
      "args": [
        "macwhisper-mcp-server"
      ],
      "env": {
        "MACWHISPER_ALLOWED_PATHS": "<MACWHISPER_ALLOWED_PATHS>"
      }
    }
  }
}
```

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

```json
{
  "servers": {
    "macwhisper-mcp-server": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "macwhisper-mcp-server"
      ],
      "env": {
        "MACWHISPER_ALLOWED_PATHS": "<MACWHISPER_ALLOWED_PATHS>"
      }
    }
  }
}
```

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

```json
{
  "mcpServers": {
    "macwhisper-mcp-server": {
      "command": "uvx",
      "args": [
        "macwhisper-mcp-server"
      ],
      "env": {
        "MACWHISPER_ALLOWED_PATHS": "<MACWHISPER_ALLOWED_PATHS>"
      }
    }
  }
}
```

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

```json
{
  "mcpServers": {
    "macwhisper-mcp-server": {
      "command": "uvx",
      "args": [
        "macwhisper-mcp-server"
      ],
      "env": {
        "MACWHISPER_ALLOWED_PATHS": "<MACWHISPER_ALLOWED_PATHS>"
      }
    }
  }
}
```

## Tools

- `transcribe_audio`: Transcribes an audio file and returns the transcript as plain text.
- `list_models`: Lists installed transcription models.
- `cancel_transcription`: Cancels the running transcription.
- `list_allowed_paths`: Returns the directories the server may read from.
- `start_watch`: Watches a folder and auto-transcribes new audio files.
- `stop_watch`: Stops the folder watcher.
- `get_watch_results`: Returns completed watch-folder transcriptions.

## FAQ

### Does audio leave my Mac?

No. The README says MacWhisper transcribes locally and no cloud APIs are used.

### What do I need?

macOS, a licensed MacWhisper with its CLI enabled, Python 3.13 and Claude Desktop.

### Can I upload audio in the chat?

No. Files must be saved in an allow-listed folder on your Mac.

### Which formats work?

m4a, mp3, mp4, mov, wav, aiff and flac.

### Is Macwhisper MCP Server free?

Yes. Macwhisper MCP Server is open source, so you can use and self-host it for free.

### How do I add Macwhisper MCP Server to Claude Code?

Run this command in your terminal: claude mcp add macwhisper-mcp-server --env MACWHISPER_ALLOWED_PATHS='<MACWHISPER_ALLOWED_PATHS>' -- uvx macwhisper-mcp-server

### What are the best alternatives to Macwhisper MCP Server?

The closest alternatives to Macwhisper MCP Server are HyperFrames by HeyGen, next-ai-draw-io, mcp-for-blender, ffmpeg-skill and cursor-talk-to-figma-mcp.

## Where this comes from

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

## Alternatives

- [HyperFrames by HeyGen](https://ai-powered.com/mcp/hyperframes-by-heygen): HyperFrames by HeyGen is a freemium MCP server for design and media.
- [next-ai-draw-io](https://ai-powered.com/mcp/next-ai-draw-io-2): next-ai-draw-io is an open-source MCP server for ai models and llm tooling.
- [mcp-for-blender](https://ai-powered.com/mcp/mcp-for-blender): mcp-for-blender is an open-source MCP server for design and media.
- [ffmpeg-skill](https://ai-powered.com/mcp/ffmpeg-skill): ffmpeg-skill is an open-source MCP server for design and media.
- [cursor-talk-to-figma-mcp](https://ai-powered.com/mcp/cursor-talk-to-figma-mcp): cursor-talk-to-figma-mcp is an open-source MCP server for design and media.
- [SSRS2](https://ai-powered.com/mcp/ssrs2): SSRS2 is an MCP server for files and documents.
- [Design Feeds](https://ai-powered.com/mcp/design-feeds): Design Feeds is a freemium MCP server for design and media.
- [Csv](https://ai-powered.com/mcp/csv): Csv is a freemium MCP server for files and documents.

---
Source: https://ai-powered.com/mcp/macwhisper-mcp-server
