dictto-mcp

MCP server for Dictto — local-first speech recognition and speaker diarization for macOS.

Connects Claude Code to your running Dictto app, giving Claude access to transcription, speaker diarization, and human-in-the-loop speaker identification tools.

Prerequisites

• Dictto macOS app installed and running
• MCP API Server enabled in Dictto: Settings > Privacy > MCP API Server
• Node.js 18+

Setup

Add to your Claude Code MCP configuration (~/.claude/claude_code_config.json):

{
  "mcpServers": {
    "dictto": {
      "command": "npx",
      "args": ["-y", "dictto-mcp"]
    }
  }
}

What it does

The MCP server acts as a bridge between Claude Code and Dictto's local API (localhost:8765). All audio processing happens on-device — nothing leaves your Mac.

Available tools (65+)

| Category | Tools | |----------|-------| | Pipeline | Run transcription + diarization, cancel, check progress, recluster | | Segments | Get, split, merge, delete, assign speakers, confirm | | Speakers | List, add, rename, merge speakers | | Known Speaker Profiles | Create, match, confirm, link, audit, manage embeddings | | Voice Analysis | Voice quality, clean regions, embedding quality, overlap detection | | Export | Full JSON export, plain text transcript | | UI | Select and play segments in the Dictto app |

Human-in-the-loop workflow

Audio file → Pipeline → Auto-labeled segments
                              ↓
              Claude reviews & confirms speakers
                              ↓
              Known Speaker profiles improve
                              ↓
              Next file needs fewer confirmations

Configuration

| Environment variable | Default | Description | |---------------------|---------|-------------| | DEVTOOLS_API_URL | http://localhost:8765/api | Dictto API endpoint |

License

MIT