Podface - AI Video Podcast Generator
AI & Desktop Application
Year
2025
Context
Final Academic Project [Group Project]
Role
Fullstack AI Engineer
Python
React
Electron.jsProject Overview
An AI-powered desktop application that humanizes digital content by generating synchronized 3D speech-driven facial animations directly from multi-speaker audio files.
The Problem
While AI can generate high-quality podcast audio, it lacks a visual component. Existing 3D animation frameworks were limited to processing a single speaker at a time, creating a high barrier to entry for automated video podcasting.
The Solution
I engineered an advanced pipeline that separates audio tracks (SpeechBrain), standardizes voices (Seed-VC), and constructs 3D meshes (MICA). I then orchestrated these conflicting Python frameworks to communicate seamlessly with a React/Electron frontend.
The Impact & Learning
Registered as an Intellectual Property (DGIP). I learned advanced environment management and asynchronous processing by successfully orchestrating heavy 3D rendering algorithms with a responsive user interface.
Live Demonstration