Meeting Notes MCP

AI-powered meeting transcription, summarization with action items, and speaker analysis using OpenAI Whisper and Qwen.

Features

Meeting Notes MCP is an MCP (Model Context Protocol) server that provides three powerful tools for processing meeting audio:

1. Transcribe Meeting ($0.08/call)

   • Convert audio to text with timestamps
   • Support for multiple languages (English, Korean, etc.)
   • Segment-based output for easy navigation

2. Summarize Meeting ($0.10/call)

   • Generate concise or detailed meeting summaries
   • Extract action items automatically
   • Powered by Qwen 3.5 LLM

3. Analyze Meeting ($0.12/call)

   • Estimate number of speakers
   • Identify key topics discussed
   • Detect overall sentiment
   • Extract important phrases

Powered By

• OpenAI Whisper (MIT License) — State-of-the-art speech recognition
• Qwen 3.5 (Apache 2.0 License) — Advanced language model for summarization and analysis
• ntriq Local AI — Secure on-premises AI inference

How to Use

Connect to Claude Desktop

Add to your Claude Desktop configuration:

{
  "mcpServers": {
    "meeting-notes": {
      "url": "https://ntriqpro--meeting-notes-mcp.apify.actor/mcp?token=YOUR_APIFY_TOKEN"
    }
  }
}

Standby Mode

Get the Standby URL from Apify and use with HTTP requests:

curl -X POST "https://ntriqpro--meeting-notes-mcp.apify.actor/mcp?token=$APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "jsonrpc": "2.0",
    "id": 1,
    "method": "tools/call",
    "params": {
      "name": "transcribe_meeting",
      "arguments": {
        "audio_url": "https://example.com/meeting.mp3"
      }
    }
  }'

API Reference

Tool: transcribe_meeting

Transcribe audio to text with timestamps.

Input:

• audio_url (string, required) — URL of audio file
• language (string, optional) — Language code (e.g., "en", "ko")

Output:

{
  "status": "success",
  "data": {
    "text": "Full meeting transcript...",
    "segments": [
      {
        "start": 0,
        "end": 5.5,
        "text": "Opening statement...",
        "confidence": 0.98
      }
    ],
    "language": "en"
  },
  "model": "whisper-1",
  "metadata": {
    "endpoint": "/audio/transcribe",
    "timestamp": "2026-03-29T10:30:00Z",
    "inputUrl": "..."
  }
}

Tool: summarize_meeting

Generate meeting summary with optional action items.

Input:

• audio_url (string, required) — URL of audio file
• summary_type (enum, default: action_items) — Type of summary:
  • "brief" — One-paragraph summary
  • "detailed" — Multi-paragraph with sections
  • "action_items" — Focus on action items and decisions

Output:

{
  "status": "success",
  "data": {
    "transcript": "Full transcript...",
    "summary": "Meeting summary...",
    "summaryType": "action_items",
    "actionItems": [
      {
        "item": "Implement new API endpoint",
        "owner": "John",
        "dueDate": "2026-04-05"
      }
    ]
  },
  "model": "qwen-3.5",
  "metadata": {
    "endpoint": "/audio/summarize",
    "timestamp": "2026-03-29T10:30:00Z",
    "summaryType": "action_items"
  }
}

Tool: analyze_meeting

Full meeting analysis with speakers, topics, sentiment.

Input:

• audio_url (string, required) — URL of audio file

Output:

{
  "status": "success",
  "data": {
    "speakersEstimated": 3,
    "topics": [
      "Product roadmap",
      "Q2 goals",
      "Team expansion"
    ],
    "sentiment": "positive",
    "keyPhrases": [
      "launch timeline",
      "resource allocation",
      "customer feedback"
    ],
    "summary": "Executive summary of the meeting..."
  },
  "model": "qwen-3.5",
  "metadata": {
    "endpoint": "/audio/analyze",
    "timestamp": "2026-03-29T10:30:00Z"
  }
}

Pricing

Pay-per-event model. Charges are deducted per successful API call:

Tool	Price
transcribe_meeting	$0.08
summarize_meeting	$0.10
analyze_meeting	$0.12

No charges for failed calls or errors.

Error Handling

All tools return consistent error format:

{
  "status": "error",
  "error": "Error message describing what went wrong"
}

Common errors:

• Timeout (>120 seconds) — "Transcription timeout (exceeded 120 seconds)"
• Invalid URL — "HTTP 404: Audio file not found"
• Network failure — "Connection refused"

Supported Audio Formats

The underlying service supports:

• MP3, WAV, FLAC, OGG
• Maximum file size: 500 MB
• Maximum duration: 24 hours

Requirements

• Node.js 18.0 or higher
• Internet connection to ai.ntriq.co.kr
• Valid Apify account for standby mode

Health Check

$curl https://ntriqpro--meeting-notes-mcp.apify.actor/health

Response:

{
  "status": "ok",
  "service": "meeting-notes-mcp",
  "version": "1.0.0",
  "timestamp": "2026-03-29T10:30:00.000Z"
}

License

This MCP server is provided under the MIT License. The underlying models use:

• OpenAI Whisper (MIT License)
• Qwen Model (Apache 2.0 License)

Support

For issues or questions:

1. Check the /health endpoint
2. Review error messages in stdout logs
3. Verify audio URL accessibility
4. Contact support@ntriq.co.kr

---

Made with by ntriq Engineering Team. Powered by Apify and Claude.