Web Lip-Sync Face Swap App or web page

Job ID: 40532536

Budget: $10 – $30 USD

I want to replicate the Wombo-style lip-sync face swap shown here: https://www.youtube.com/watch?v=YrgXUP1Yj_Y, but delivered as a browser-based experience. The interface should stay simple and minimalistic—users land on a clean page, upload a photo, pick or record a short clip, and instantly see the face-swapped, lip-synced result that they can download or share.

My biggest concern is cost: most commercial APIs charge per swap and the fees add up fast. I’m open to self-hosting open-source models (e.g., DeepFaceLab, FaceSwap, or face-tracking libraries built on TensorFlow/PyTorch) or any lightweight inference service that keeps runtime costs close to zero. When you respond, outline exactly how you would avoid or drastically reduce per-request fees—model choice, optimisation method, and hosting stack.

Core features I need you to deliver:
• Ability for users to upload their own photos
• Real-time or near real-time face swapping with lip-sync to audio/video clips
• One-click save and share options (direct download link, shareable URL, or social APIs)

Please include:
– A working web prototype (React, Vue, or plain JS—your choice)
– Server or edge-function code handling the swap pipeline
– Deployment instructions and documentation explaining the cost-saving approach, so I can run it cheaply on my own server or a low-cost cloud instance

If you have a demo or previous work that proves you can run face-swap inference without pricey third-party calls, that will move you to the top of the list.