Search papers, labs, and topics across Lattice.
3
0
5
11
Autoregressive generative models, previously unsuitable for real-time target speaker extraction, can now achieve offline-level performance in streaming scenarios thanks to a novel chunk-wise splicing technique.
Seedance 2.0 leapfrogs existing models by unifying multi-modal inputs (text, image, audio, video) into a single architecture for generating high-quality, longer-duration audio-video content.
DreamID-Omni lets you precisely control multiple character identities and voice timbres in generated audio-video, even outperforming proprietary models.
