Project Overview
Duix-Avatar (formerly known as HeyGem) is a free and open-source project released by Silicon Intelligence, specifically designed for Windows systems, supporting fully offline video synthesis and digital human cloning.
Core Features
- Precise Appearance Cloning: Uses advanced AI algorithms to capture facial features with high precision, building realistic virtual models
- Voice Cloning: Accurately clones voice characteristics, reproducing subtle features of speech with support for various voice parameter settings
- Text/Voice Driven: Through natural language processing technology, converts text into natural and fluent speech to drive virtual avatars
- Efficient Video Synthesis: Intelligently optimizes audio-video synchronization effects, achieving natural and smooth lip-sync
- Multi-language Support: Supports 8 languages including English, Japanese, Korean, Chinese, French, German, Arabic, and Spanish
Technical Advantages
- Fully Offline Operation: No internet connection required, effectively protecting user privacy and avoiding potential data leakage risks
- User-Friendly: Clean and intuitive interface, easy to use even for beginners without technical background
- Multiple Model Support: Supports importing multiple models and managing them through one-click startup packages
- Cross-Platform Compatibility: Supports NVIDIA 50 series graphics cards, Ubuntu version released
Application Scenarios
- Virtual anchor and voice-over video production
- Enterprise digital human customer service
- Education and training virtual instructors
- Content creation and marketing
- Personal IP digitization
Technical Architecture
- ASR: Automatic speech recognition technology based on fun-asr
- TTS: Speech synthesis technology based on fish-speech-ziming
- Computer Vision: Used for facial recognition and lip movement analysis, ensuring virtual avatar lip movements match voice content
Open Source: https://github.com/duixcom/Duix-Avatar
Official Website: https://duix.com
Former Name: HeyGem
Loading...