{
  "markdown": "# Whisper MCP Server\n\nA lightweight MCP (Model Context Protocol) server for local audio transcription using [whisper.cpp](https://github.com/ggerganov/whisper.cpp). There are [several Whisper MCP implementations](https://github.com/search?q=whisper+mcp&type=repositories) out there. This one is minimal and pairs with [apple-voice-memo-mcp](https://github.com/jwulff/apple-voice-memo-mcp) for a complete voice memo workflow.\n\n## Features\n\n- **Local transcription** - All processing happens on your machine\n- **Multiple models** - Choose from tiny, base, small, medium, or large models\n- **Various formats** - Supports wav, mp3, m4a, and other audio formats\n- **Timestamps** - Get transcriptions with or without timestamps\n\n## Requirements\n\n- macOS (tested on Apple Silicon)\n- Node.js 18+\n- whisper-cpp: `brew install whisper-cpp`\n- ffmpeg: `brew install ffmpeg`\n\n## Installation\n\n```bash\nnpm install -g whisper-mcp\n```\n\nOr run directly:\n\n```bash\nnpx whisper-mcp\n```\n\n## Configuration\n\n### Claude Desktop\n\nAdd to your Claude Desktop config file:\n\n**macOS**: `~/Library/Application Support/Claude/claude_desktop_config.json`\n\n```json\n{\n  \"mcpServers\": {\n    \"whisper-mcp\": {\n      \"command\": \"npx\",\n      \"args\": [\"-y\", \"whisper-mcp\"]\n    }\n  }\n}\n```\n\nAfter editing, restart Claude Desktop.\n\n### Claude Code (CLI)\n\nFor Claude Code, add to your project's `.mcp.json` file:\n\n```json\n{\n  \"mcpServers\": {\n    \"whisper-mcp\": {\n      \"command\": \"npx\",\n      \"args\": [\"-y\", \"whisper-mcp\"]\n    }\n  }\n}\n```\n\nOr for user-wide configuration, add to `~/.claude/settings.json`:\n\n```json\n{\n  \"mcpServers\": {\n    \"whisper-mcp\": {\n      \"command\": \"npx\",\n      \"args\": [\"-y\", \"whisper-mcp\"]\n    }\n  }\n}\n```\n\n**Tip**: Use `/mcp` in Claude Code to verify the server is connected.\n\n### Local Development Setup\n\nIf running from source instead of npm:\n\n```json\n{\n  \"mcpServers\": {\n    \"whisper-mcp\": {\n      \"command\": \"node\",\n      \"args\": [\"/path/to/whisper-mcp/dist/index.js\"]\n    }\n  }\n}\n```\n\n### With Apple Voice Memos MCP\n\nFor a complete voice memo workflow, use alongside apple-voice-memo-mcp:\n\n```json\n{\n  \"mcpServers\": {\n    \"apple-voice-memo-mcp\": {\n      \"command\": \"npx\",\n      \"args\": [\"-y\", \"apple-voice-memo-mcp\"]\n    },\n    \"whisper-mcp\": {\n      \"command\": \"npx\",\n      \"args\": [\"-y\", \"whisper-mcp\"]\n    }\n  }\n}\n```\n\n## MCP Tools\n\n### `transcribe_audio`\n\nTranscribe an audio file using Whisper.\n\n**Parameters:**\n- `file_path` (required): Absolute path to the audio file\n- `model` (optional): Model to use (tiny.en, base.en, small.en, medium.en, large). Default: base.en\n- `language` (optional): Language code. Default: en\n- `output_format` (optional): text, timestamps, or json. Default: text\n\n**Example:**\n```json\n{\n  \"file_path\": \"/path/to/audio.m4a\",\n  \"model\": \"medium.en\",\n  \"output_format\": \"timestamps\"\n}\n```\n\n### `list_whisper_models`\n\nList available Whisper models and their download status.\n\n**Returns:**\n```json\n{\n  \"models\": [\n    {\n      \"name\": \"base.en\",\n      \"size\": \"142 MB\",\n      \"downloaded\": true,\n      \"path\": \"/Users/you/.whisper/ggml-base.en.bin\"\n    }\n  ]\n}\n```\n\n### `download_whisper_model`\n\nDownload a Whisper model for local use.\n\n**Parameters:**\n- `model` (required): Model to download (tiny.en, base.en, small.en, medium.en, large)\n\n## Models\n\n| Model | Size | Speed | Quality |\n|-------|------|-------|---------|\n| tiny.en | 75 MB | Fastest | Basic |\n| base.en | 142 MB | Fast | Good |\n| small.en | 466 MB | Medium | Better |\n| medium.en | 1.5 GB | Slow | Great |\n| large | 2.9 GB | Slowest | Best |\n\nModels are stored in `~/.whisper/`.\n\n## Workflow Example\n\n1. List your voice memos: `list_voice_memos`\n2. Get audio path: `get_audio` with memo ID\n3. Transcribe: `transcribe_audio` with the file path\n4. Save to your vault\n\n## Development\n\n```bash\n# Clone and install\ngit clone https://github.com/jwulff/whisper-mcp.git\ncd whisper-mcp\nnpm install\n\n# Build\nnpm run build\n\n# Test with MCP inspector\nnpm run inspector\n```\n\n## License\n\nMIT\n",
  "bytes": 3963,
  "sha": "188d875b2138e99572283af1224a78cb9ac7663b298eb7d58d8ee11c0bc34780",
  "repo_slug": "jwulff/whisper-mcp",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/mcp_io_github_jwulff_whisper_mcp_24d9ed08/readme"
}