Creating NPCs With Generative AI and OpenUSD | Omniverse Live
Вставка
- Опубліковано 25 лип 2023
- Join the Convai team as we explore how to add AI characters to your virtual worlds in NVIDIA Omniverse using Universal Scene Description (OpenUSD) and Generative AI.
All upcoming livestreams: www.addevent.com/calendar/ae4... - Розваги
An old question, when will convai support Chinese voice quizzes and Chinese knowledge banks?
@@carmenkarmankong3062 Thanks, I have tried it. To be honest, the accuracy of Chinese speech recognition is a bit poor. In addition, can it support simultaneous multi-language model recognition? Such as Chinese and English. Now if I say something (你是 CHATGPT么?), there will be very strange results.
The knowledge bank feature is right now english only but the character will reply in Chinese. We are working in multilingual knowledge bank but that will take some time.
How do we get access to ACE?
Is it possible to upload a 2D picture of a real person to convert it to the avatar, also can we customize the background environment?
Yes using NeRF
I prefer to use voice actors instead of A.I. voice generation, as it provides a more authentic and engaging experience. Additionally, utilizing motion capture (mocap) actions instead of preset animations would be more practical and suitable for my specific needs. I would like to know if these options are feasible for implementation.
Stop concern trolling. It isn't possible to record the endless permutations of real voices for the use cases ACE provides.
O.o
Literally no difference between AI voice generation and voice actors. Both do the same thing to the same effect and both are just as artificial as each other though voice actors are slightly higher quality for now.
The amount of data to remember conversations for millions of people seems unrealistic.
Thays exactly why this technology is so potentially revolutionary, it enables experiences that literally can't be pre-recorded!
More NVIDIA H100 is Better 😍😋🤖🦾🦾🦾🦾