Two new audio models blur reality: To show off its new lip-syncing model, AI startup Tavus turned a monologue from HBO’s The White Lotus into a rant about LLMs. Hummingbird-0 lets you transform your subject’s mouth movements using a short reference clip and audio of your choice. Meanwhile, a startup called Rime just unveiled Arcana, which can capture the “nuances of real human speech,” including accents, vocal stumbles, and more, with unprecedented realism.
https://www.tavus.io/post/introducing-hummingbird-0-a-leap-in-lip-sync