Slack Voice MCP Server

AI-powered Slack integration with text-to-speech capabilities using OpenAI GPT-4 and voice synthesis.

Features

• 🤖 AI Summarization: Intelligent message summaries using OpenAI GPT-4o-mini
• 📱 Slack Integration: Retrieve unread messages from channels and DMs
• 🔊 Text-to-Speech: Multiple TTS options (OpenAI, Piper, System)
• 🖥️ MCP Protocol: Seamless integration with Claude Desktop and other AI tools
• ⚡ Real-time: Fast message processing and voice synthesis

Quick Start

Prerequisites

• Node.js 18+
• Claude Desktop application (or other MCP-compatible AI tool)
• Slack workspace access
• OpenAI API key

Installation

1. Clone and setup:

git clone https://github.com/karl-frascari/slack-voice-mcp.git
cd slack-voice-mcp
npm install
npm run build

2. Optional: Install Piper TTS for offline voice synthesis:

Download Piper TTS files and extract to piper/ directory:

• Download from: Piper TTS Releases
• Extract piper.exe, voice models, and espeak-ng-data/ to piper/ folder
• Recommended voice: en_US-libritts-high.onnx

3. Configure environment:

cp .env.example .env
# Edit .env and add your tokens and API keys

4. Configure Claude Desktop: Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "slack-voice": {
      "command": "node",
      "args": ["/absolute/path/to/slack-voice/dist/index.js"],
      "env": {
        "SLACK_USER_TOKEN": "xoxp-your-slack-user-token",
        "SLACK_BOT_TOKEN": "xoxb-your-slack-bot-token",
        "OPENAI_API_KEY": "sk-proj-your-openai-api-key",
        "TTS_PROVIDER": "openai",
        "NODE_ENV": "production"
      }
    }
  }
}

Configuration

Environment Variables

Create a .env file based on .env.example:

# Required
SLACK_USER_TOKEN=xoxp-your-slack-user-token
SLACK_BOT_TOKEN=xoxb-your-slack-bot-token
OPENAI_API_KEY=sk-proj-your-openai-api-key

# Optional
TTS_PROVIDER=openai
MCP_SERVER_NAME=slack-voice-mcp
NODE_ENV=production
LOG_LEVEL=info

Slack Token Setup

1. Visit api.slack.com/apps
2. Create new app or use existing
3. Go to OAuth & Permissions
4. Add these scopes:
   • channels:read
   • channels:history
   • groups:read
   • groups:history
   • im:read
   • im:history
   • mpim:read
   • mpim:history
5. Install app to workspace
6. Copy User OAuth Token (starts with xoxp-)

OpenAI API Key

1. Visit platform.openai.com
2. Create an API key
3. Add to your .env file

TTS Providers

OpenAI TTS (Recommended)

High-quality cloud-based text-to-speech with multiple voice options.

Features:

• 🎭 Multiple Voices: alloy, echo, fable, onyx, nova, shimmer
• 🌐 High Quality: Natural-sounding speech synthesis
• 💰 Paid Service: Requires OpenAI API key

Piper TTS

High-quality neural text-to-speech that runs completely offline.

Features:

• 🔒 Completely Offline: No internet connection required
• ⚡ Fast: Optimized for real-time synthesis
• 🎵 Natural: LibriTTS-trained voice model

System TTS

Built-in operating system text-to-speech as fallback.

Supported Platforms:

• Windows: Microsoft Speech API
• macOS: Built-in say command
• Linux: System TTS

Usage

Available MCP Tools

1. get_summarize_and_speak - Complete workflow: get messages, summarize with AI, and speak
2. get_available_voices - List available TTS voices

Example Usage in Claude Desktop

Please get my Slack summary and read it to me.

Claude will use the MCP server to:

1. Fetch your unread Slack messages
2. Generate an intelligent summary using OpenAI GPT-4o-mini
3. Speak the summary using your configured TTS provider

Architecture

Claude Desktop (or other AI tool)
    ↕ MCP Protocol
Slack Voice MCP Server
    ├── Slack Service (Message Retrieval)
    ├── Summarization Service (OpenAI GPT-4o-mini)
    └── TTS Service (Voice Output)
        ├── OpenAI TTS
        ├── Piper TTS (Local)
        └── System TTS (Fallback)

Development

Build and Run

# Install dependencies
npm install

# Build TypeScript
npm run build

# Run in development mode
npm run dev

# Start production server
npm start

# Lint and format
npm run check
npm run fix

Project Structure

src/
├── index.ts              # MCP server entry point
├── services/
│   ├── slack.ts          # Slack API integration
│   ├── summarization.ts  # OpenAI summarization
│   └── tts.ts           # Text-to-speech providers
└── types/               # TypeScript type definitions

Troubleshooting

Common Issues

MCP connection errors:

• Ensure absolute paths in Claude Desktop config
• Check environment variables are set correctly
• Restart Claude Desktop after config changes

Slack token errors:

• Verify token starts with xoxp- (User Token)
• Check token has required scopes
• Ensure workspace access permissions

OpenAI API errors:

• Verify API key is valid and has credits
• Check internet connection
• Ensure API key has proper permissions

Audio not working:

• Try different TTS providers in order: OpenAI → Piper → System
• Check system audio settings
• For WSL users: use TTS_PROVIDER=openai

Security

• Environment Variables: Sensitive tokens stored in .env (not committed)
• Minimal Permissions: Only requests necessary Slack scopes
• No Persistent Storage: Messages processed in memory only
• Secure Communication: Uses official Slack and OpenAI APIs

License

MIT