Called the Omniverse Avatar Cloud Engine (ACE), the new toolkit will use the processing power of Nvidia’s new graphics accelerators to generate chatbot or avatar characters for use in games, or to render with the “professionalism” of a news anchor any information desired by the person behind the keyboard.
According to Nvidia, the combination of artificial intelligence models and services will help developers quickly create virtual characters that don’t require the massive computing power available with the PCs they use – a startup can produce an avatar as convincing as one from a giant corporation by tapping into Nvidia ACE services, made available with cloud processing technologies.
Nvidia ACE brings together several existing software kits and platforms, combining computer vision techniques for object recognition and computer animations designed to convincingly render facial expressions. The verbal interaction part is assisted with natural language models and advanced speech recognition.
It will probably be a long time yet, despite Nvidia’s promises, before you can speak to an ACE-based avatar in a video game or at the entrance to the hotel where you’re looking for accommodation, sounding as convincing as a real person. And while NVIDIA claims that its AI is “on track” to pass the Turing test, clearly that goal is still a long way off, at least with publicly available software, with the demo video showing that the avatar created has at best the qualities of a presenter reading a pre-prepared text.