Search papers, labs, and topics across Lattice.
4
0
5
Natural-language instructions can now handle both zero-shot speech synthesis and surgical acoustic editing within a single unified model, operating at 4-step distilled inference speeds without classifier-free guidance.
HoliDubber revolutionizes video dubbing by seamlessly integrating speech and sound effects from a single text prompt, outperforming traditional methods in quality and synchronization.
Current audio editing models are failing spectacularly, with an Exact Match Rate below 5% in complex tasks, exposing a critical need for improvement.
Online reinforcement learning with large audio language model rewards catapults text-to-audio generation to a new state-of-the-art, even with a relatively small 470M parameter model.
