Audio RAG

Transcribes audio, stores transcripts in a vector database, and enables natural-language search across audio content.

typescript

GitHub Stars

typescript

Language

6 months ago

First Indexed

3 months ago

Catalog Refreshed

Documentation & install

Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
  "mcpServers": {
    "matheusslg-mcp-audio-rag": {
      "command": "npx",
      "args": [
        "tsx",
        "/full/path/to/mcp-audio-rag/src/server.ts"
      ],
      "env": {
        "SUPABASE_URL": "https://your-project.supabase.co",
        "GEMINI_API_KEY": "your-key",
        "SUPABASE_SERVICE_KEY": "your-service-role-key"
      }
    }
  }
}

The MCP Audio RAG server lets you transcribe audio files, store the transcripts, and search them with natural language queries. It enables you to build a searchable knowledge base from meetings, podcasts, lectures, and other audio, then ask Claude questions to discover key topics, mentions, and summaries across your collection.

How to use

You interact with the MCP server through your MCP client (Claude or other compatible clients). You can transcribe audio files, search across transcripts with natural language, and retrieve full transcripts or summarized insights. Use explicit commands to drive transcription, searches, and library management, and rely on the server to return relevant results with highlighted snippets.

How to install

Prerequisites you need before starting are Node.js 18 or newer, access to a Gemini API key, and a Supabase account.

git clone https://github.com/matheusslg/mcp-audio-rag.git
cd mcp-audio-rag
npm install

Additional setup and configuration

Prepare your database and API keys to enable transcription and search capabilities.

cp .env.example .env

Edit the environment file to insert your API keys and database details.

GEMINI_API_KEY=your-key-here
SUPABASE_URL=https://your-project.supabase.co
SUPABASE_SERVICE_KEY=your-service-role-key

Claude integration and runtime configuration

You connect the MCP server to Claude so you can run transcriptions and queries from Claude’s interface.

{
  "mcpServers": {
    "audio-rag": {
      "command": "npx",
      "args": ["tsx", "/full/path/to/mcp-audio-rag/src/server.ts"],
      "env": {
        "GEMINI_API_KEY": "your-key",
        "SUPABASE_URL": "https://your-project.supabase.co",
        "SUPABASE_SERVICE_KEY": "your-service-role-key"
      }
    }
  }
}

Usage examples

Transcribe an audio file using the integrated Gemini model, then search for mentions of a topic across your transcripts.

Commands you can instruct Claude with include transcription requests, searches for terms, and operations to list or summarize transcripts.

Security and troubleshooting

Ensure the environment variables are correctly supplied and kept secure. If you see missing environment variable errors, double-check the .env file and Claude’s config to confirm all three keys are present.

If you encounter slow transcription, consider using a lighter Gemini model for bulk processing.

Supabase access issues often stem from using the anon key instead of the service_role key. Verify you are using the correct key type.

Available tools

ingest_audio

Transcribe and store an audio file into the database with its transcript.

search_transcripts

Search transcripts using natural language queries and retrieve matching segments.

list_transcripts

List all transcribed audio files saved in the database.

get_full_transcript

Retrieve the complete transcript for a specific audio file.

summarize_audio

Generate an AI summary of a transcript for quick insights.

delete_transcript

Remove a transcribed file from the database.

Built by

VeilStrat

AI signals for GTM teams