Search papers, labs, and topics across Lattice.
Affiliation:
2
0
5
HiCoDiT achieves superior audio-visual alignment by harnessing the hierarchical nature of speech tokens, outperforming traditional VTS methods in both fidelity and expressiveness.
