Sealed Rose

High-precision multi-modal AI deepfake detection, voice clone verification, and creator authenticity platform protecting digital identities in real time.

Comment

Inspiration

What it does

💡 Inspiration

As generative AI models (Sora, Flux, Midjourney, ElevenLabs) advance, digital deception and romance/catfish scams have exploded. Deepfake video calls and synthetic voices now routinely bypass traditional human scrutiny. We built Sealed Rose to provide an immediate, accessible, and uncompromising forensics suite that allows anyone—from independent creators to dating app users—to verify the authenticity of media in seconds.

🔍 What it does

Sealed Rose is a full-spectrum digital authenticity and forensics suite:

  • Video Forensics & Deepfake Detection: Inspects video frames for generative artifacts, frequency anomalies, facial warping, and temporal inconsistencies.
  • Client-Side Keyframe Sampling: Employs an in-browser sampling pipeline capable of analyzing high-resolution multi-gigabyte videos in seconds without crashing servers or requiring massive bandwidth.
  • Image AI Inspection: Detects synthetic textures, GAN/Diffusion patterns, and generative signatures across uploaded photos.
  • Voice Clone Verification: Analyzes audio spectrograms to detect vocal cloning and synthetic voice synthesis.
  • OSINT & Anti-Catfish Intelligence: Cross-checks usernames, phone footprints, and leak databases to protect individuals against extortion and impersonation.

🛠️ How we built it

  • Frontend & App Architecture: Built with Next.js 15 (App Router), React 19, and Tailwind CSS for a responsive, dark-brass luxury aesthetic.
  • Backend & Database: Powered by PostgreSQL hosted on Railway, managed via Drizzle ORM with automated migrations.
  • Client-Side Video Forensics: Implemented custom HTML5 <video> and <canvas> sampling engines in pure TypeScript to extract and compress native-resolution keyframes locally before dispatching to verification pipelines.
  • Authentication: Integrated multi-provider OAuth, including Sign in with Apple (with dynamic ES256 JWT signing) and Google One Tap.
  • Edge Routing & Performance: Optimized with Cloudflare DNS, CDN caching, and automated search engine IndexNow broadcasting.

🧗 Challenges we ran into

Handling large video files was our biggest bottleneck. When users attempted to upload 1GB+ 1080p videos from mobile devices, traditional server routes struggled with timeouts and payloads. We engineered a dual-action client-side solution: extracting mathematical keyframes in browser memory in under 3 seconds, paired with a dedicated Quick Pass forensic pipeline for heavy enterprise analysis.

🏆 Accomplishments that we're proud of

  • Shipped directly to production serving real-time traffic across 20+ countries.
  • Successfully verified over 280+ media files with sub-3-second average turnaround.
  • Zero-dependency client-side frame sampling that works seamlessly on mobile Safari and Chrome.

🔮 What's next for Sealed Rose

  • Expanding our Chrome Extension to auto-flag deepfakes on X, Instagram, and dating web apps.
  • Rolling out continuous live video stream analysis for real-time video call protection. ## How I built it

Challenges I ran into

Accomplishments that I'm proud of

What I learned

What's next for Sealed Rose

Share this project:

Updates