Experiment — Alpha Quality

This project is an early-stage experiment. It may break, change significantly, or be abandoned. Use it for inspiration and fun, not production workloads.

EchoDeck

Experiment

Give Claude your voice

An MCP server that connects to Claude and converts PowerPoint presentations into narrated MP4 videos. Claude generates the deck with speaker notes, EchoDeck renders it into a finished video using ElevenLabs or Google Gemini TTS — no video editing or microphone required.

C#MCPElevenLabsGoogle GeminiDocker
View Source on GitHub Deploy on Railway

What It Does

  • Ask Claude to generate a presentation — it writes the slides and speaker notes
  • EchoDeck converts the PPTX into a narrated MP4 automatically
  • Choose your voice: ElevenLabs or Google Gemini TTS
  • Configurable resolution: 720p, 1080p, or 4K
  • Configurable slide transitions
  • Web UI with real-time progress tracking
  • One-click Railway deployment with auto-generated MCP auth keys

Caveats

  • This is an alpha-quality experiment — expect rough edges
  • Requires an ElevenLabs or Google Gemini API key
  • Tool call limits on longer videos may require manual "continue" prompts
  • Generated videos are best for sharing and demos, not broadcast production

Quick Start

// step 1 — get a voice

Get an ElevenLabs API key

elevenlabs.io →

// step 2 — deploy your server

Deploy EchoDeck on Railway — set ELEVENLABS_API_KEY when prompted

Deploy on Railway

// step 3 — connect to claude

Add your MCP URL as a Custom Integration in Claude Desktop

https://{your-railway-url}/mcp

How It Works

A real example, start to finish.

  1. 1

    Ask Claude to build a deck

    "Can you make me a powerpoint on some cool new topic, make it 2 slides and include speaker notes"
  2. 2

    Claude builds the entire presentation

    It picks a topic, writes the slides, adds speaker notes, and hands you a downloadable .pptx file.

  3. 3

    Ask Claude to generate a video

    "Great now generate this in Bella's voice"

    Bella is one of ElevenLabs' default voices — no setup needed. You can use any voice from your ElevenLabs account, including a cloned one.

  4. 4

    Video is ready

    1080p, with crossfade transitions between slides and your chosen voice reading the speaker notes.

  5. 5

    The output

    Generated entirely by Claude + EchoDeck. No recording. No editing. No microphone.