Build a voice first AI agent that can see and respond to a user in real time.
Create a talking digital human character from a single photo.
Add persona memory so an AI character remembers past conversations.
Experiment with retrieval augmented generation for a voice assistant.
| dsd2077/cyberverse | paddlepaddle/paddlets | truegunsmithfence/hentaihunter | |
|---|---|---|---|
| Stars | 545 | 547 | 542 |
| Language | Python | Python | Python |
| Last pushed | — | 2026-04-17 | — |
| Maintenance | — | Maintained | — |
| Setup difficulty | hard | moderate | easy |
| Complexity | 4/5 | 3/5 | 2/5 |
| Audience | developer | data | general |
Figures from each repo's GitHub metadata at analysis time.
Requires setting up WebRTC infrastructure and API keys for the underlying AI services.
CyberVerse is an open source platform for building real time voice and video AI agents. It uses WebRTC, the same browser technology behind video calls, to stream audio and video between a user and an AI character with low delay. The project's stated goal is to let a single photo become the starting point for a talking digital character that can see, hear, and respond to a person in real time. Under the hood it combines persona memory, so the character remembers details about who it is and who it is talking to, with tools and retrieval augmented generation, which means the AI can pull in extra information beyond its built in knowledge before answering. An optional digital human video layer generates a moving, speaking face to go along with the voice output, rather than a static avatar or text only chat window. The project is aimed at people who want to build their own AI companions, assistants, or characters, similar in spirit to fan made AI chat tools but with an emphasis on live voice and video rather than typed messages. It is written mostly in Python and supports multiple languages in its documentation, including English, Chinese, Japanese, and Korean, suggesting an international user base. CyberVerse is licensed under the GNU General Public License version 3, a copyleft license that requires any modified versions distributed to others to also be released as open source under the same terms. The README frames the project with emotional language about recreating the voice and presence of someone the user has lost, so anyone evaluating it should be aware this is explicitly part of its stated use case, alongside more general purpose voice agent building.
An open source platform for building real time voice and video AI agents and digital human characters using WebRTC.
Mainly Python. The stack also includes Python, WebRTC, RAG.
You can use and modify this freely, but if you distribute a modified version, you must also release its source code under the same license.
Setup difficulty is rated hard, with roughly 1h+ to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.