From ChatGPT to MLC LLM, a local chatbot that can run on Google Bard and mobile phones, there are many ways to do text chat with large language models. AI’s next frontier is bringing LLM’s powers to her NPCs (Non-Player Characters) in-game. There, you can have broad and open conversations instead of platitudes.
In his Computex 2023 keynote, Nvidia CEO Jensen Huang talks about bringing game characters to life using natural language dialogue, speech-to-expression, and text-to-speech/speech-to-speech. Announcing ACE for Games, an AI model foundry service designed to inspire. -Text function. Huang describes his demo of a game in which his NPC, named Jin, who runs a ramen shop, interacts with a human player who asks spoken questions and receives authentic-looking answers that match the NPC’s backstory. I showed off.
In the demo, a gamer (named Kai) walks into Jin’s ramen shop, asks (via voice) how he’s doing, and converses with him about the fact that the area has a high crime rate. When Kai asked if he could help, Jin said, “If you want to do something about this, I’ve heard rumors that powerful crime lord Kumon Aoki is causing all sorts of chaos in the city. He may be the root of this violence.” . ” When Kai asks where Aoki is, Jin answers him and begins searching for the user.
“AI will not only contribute to the redification and synthesis of the environment, but will also give animation to the characters,” Huang said. “AI will be a huge part of the future of video games.”
Nvidia ACE for Games provides fast access to three different existing components. The first, Nvidia NeMo, is an AI framework for LLM training and deployment that includes NeMo Guardrails designed to prevent inappropriate or “unsafe” AI conversations. Presumably this prevents the NPC from responding to inappropriate and off-topic prompts from the user. Guardrails also provide security, prompting users and would-be injectors to “prison break” Attack bots and make them do bad things.
Nvidia Riva is the company’s text-to-speech/text-to-speech solution. In the ACE for games workflow, gamers ask questions through the microphone and Riva converts them to text and sends them to LLM. LLM then generates a text response, which Riva converts back into speech for the user to hear. Of course, the game also expects the response to be displayed in text. You can try out Nvidia Riva’s text-to-speech and text-to-speech features for yourself. company site.
Nvidia Omniverse Audio2Face provides the final step in the ACE workflow for games, allowing characters to have facial expressions that match what they are saying. The company is currently offering this product in beta. try here.
This demo, called Kairos, combineAn in-game AI startup that is part of Nvidia’s Inception Program We connect startups with venture capital. The company’s site offers a toolset that allows game developers to build his lifelike NPC with a complex backstory.
