Apple Voice Memo

This MCP server provides programmatic access to Apple Voice Memos on macOS, enabling you to list memos, inspect details, retrieve audio, and obtain or generate transcripts for your recordings. It integrates with Claude and other AI assistants to make your voice memos searchable and usable in automated workflows.

How to use

You connect to this server using an MCP client or integration. With it, you can list your voice memos and filter them, request detailed metadata for a specific memo, fetch the audio file path or base64 data, extract existing transcripts embedded in the audio, and generate new transcripts when needed. Transcriptions leverage Apple's SFSpeechRecognizer where available. Use the server to automate querying your Voice Memos data from your preferred AI assistant or automation tool.

How to install

Prerequisites you need before installing the server:

• macOS Sonoma (14.0) or later; Sequoia recommended for transcription
• Node.js 18 or newer
• Full Disk Access permission to read Voice Memos data
• Voice Memos app must have been opened at least once

Install the MCP server using npm or npx as shown:

npm install -g apple-voice-memo-mcp

Or run directly with npx:

npx apple-voice-memo-mcp

Configuration

To connect from an MCP client, configure the client to start the server via a simple command. Below are examples that you can use in your client configuration.

{
  "mcpServers": {
    "apple-voice-memo-mcp": {
      "command": "npx",
      "args": ["apple-voice-memo-mcp"]
    }
  }
}

Security and permissions

Grant the necessary permissions so the server can access your Voice Memos data and run transcription when needed.

• Full Disk Access is required to read the Voice Memos database
• Speech recognition for transcription is required only if you use the transcribe memo tool

Troubleshooting

If you encounter issues starting or using the server, check the following common scenarios.

• Voice Memos database not found: ensure the Voice Memos app has been opened at least once and Full Disk Access is granted.
• No transcript available: macOS Sequoia+ can auto-transcribe when opened; otherwise use transcribe_memo.
• Native module binding errors: run npm rebuild better-sqlite3

Notes and development

Transcripts are stored inside the .m4a audio files using a custom MPEG-4 atom named tsrp. The MCP server reads these embedded transcripts directly from the audio files; there are no separate transcript files.

Development and testing

If you are developing locally, you can run the server from source using the standard Node workflow and test with an MCP inspector.

# Clone the repository
git clone https://github.com/jwulff/apple-voice-memo-mcp.git
cd apple-voice-memo-mcp

# Install dependencies
npm install

# Build
npm run build

# Run in development
npm run dev

# Test with MCP inspector
npm run inspector

Glossary and endpoints

This server exposes a set of tools to interact with Voice Memos data through MCP. You can list memos, retrieve memo details, get audio, obtain transcripts, and transcribe memos using Apple's recognition capabilities.

Available tools

list_voice_memos

List all voice memos with metadata including id, title, date, duration, and transcript availability.

get_voice_memo

Get detailed metadata for a specific memo by its ID.

get_audio

Retrieve the audio file path or base64-encoded audio data for a memo.

get_transcript

Extract the transcript embedded in a memo's audio file.

transcribe_memo

Transcribe a memo using Apple's SFSpeechRecognizer.