VOD Profiler

A full-stack DMCA risk scanner for streamers. Submit a video file or URL and the app fingerprints the audio via ACRCloud, returning a breakdown of identified copyrighted tracks with timestamps, confidence scores, and DMCA risk flags.

Live: vod-profiler.vercel.app

Architecture

User submits video file or URL
              |
              v
Chunked Upload (4MB chunks via frontend JS)
              |
              v
Express.js API (Vercel Serverless Functions)
              |
              v
ACRCloud Audio Fingerprinting API
              |
              v
Parsed Results: track title, artist, timestamps, confidence score, DMCA risk flag

Tech Stack

• Backend: Node.js, Express.js
• Deployment: Vercel (Serverless Functions)
• Audio Fingerprinting: ACRCloud REST API (HMAC-SHA1 auth)
• Frontend: HTML, Vanilla JavaScript

What This Project Demonstrates

Chunked Streaming Architecture: Vercel serverless functions have a 4.5MB request payload limit. The frontend splits files into 4MB chunks and POSTs each independently. The server stores chunks in /tmp and reassembles them once all chunks arrive — enabling files 10x the default payload limit to process without timeout.

Third-Party API Integration: ACRCloud uses HMAC-SHA1 signed requests for authentication. The signature is computed client-side and Base64-encoded on each request. The app parses the structured response to extract timestamps, match confidence scores, and map segments to DMCA risk levels.

Serverless Deployment: The entire backend runs as Vercel Serverless Functions. vercel.json configures routing, memory limits, and timeout settings — no server management required.

Screenshots

File Upload and Scan Results

How to Run Locally

Prerequisites: Node.js installed, ACRCloud account for API credentials.

# Clone and install dependencies
git clone https://github.com/jpgoreczky/VOD-Profiler.git
cd VOD-Profiler
npm install

# Configure environment variables
cp .env.example .env
# Edit .env and add your ACRCloud credentials:
# ACRCLOUD_HOST=identify-eu-west-1.acrcloud.com
# ACRCLOUD_ACCESS_KEY=your_access_key
# ACRCLOUD_ACCESS_SECRET=your_access_secret

# Start the development server
npm run dev

Open http://localhost:3000 in your browser.

Running Tests

npm test

Tests cover the ACRCloud signature builder and the response parser.

API Reference

POST /api/upload

Accepts one chunk of a large file upload. Returns intermediate progress or final results on the last chunk.

POST /api/recognize

Single-shot recognition for small audio clips (4MB or less).

Result object:

{
  "trackTitle": "Blinding Lights",
  "artist": "The Weeknd",
  "timestampStart": "00:14:32",
  "timestampEnd": "00:17:45",
  "confidenceScore": 92,
  "dmcaRisk": "HIGH"
}

DMCA Risk Levels:

Level	ACRCloud Confidence
HIGH	80% and above
MEDIUM	50 to 79%
LOW	Below 50%

Project Structure

VOD-Profiler/
├── api/
│   ├── index.js          # Local dev Express server
│   ├── upload.js         # Chunked upload + ACRCloud scan endpoint
│   └── recognize.js      # Single-shot recognition endpoint
├── src/services/
│   ├── acrcloud.js       # ACRCloud API client (auth + HTTP)
│   └── parser.js         # ACRCloud response to flagged-segment mapper
├── public/
│   ├── index.html        # Frontend UI
│   └── app.js            # Chunked upload logic + results rendering
├── tests/
│   ├── acrcloud.test.js  # Signature generation unit tests
│   └── parser.test.js    # Response parsing unit tests
├── .env.example          # Environment variable template
├── vercel.json           # Vercel deployment configuration
└── package.json