Search papers, labs, and topics across Lattice.
Fudan University
1
0
2
HiCoDiT achieves superior audio-visual alignment by harnessing the hierarchical nature of speech tokens, outperforming traditional VTS methods in both fidelity and expressiveness.
