Search papers, labs, and topics across Lattice.
1
0
3
Get state-of-the-art spoken QA performance by adding lightweight speech modules to frozen VL models and training on synthetically generated speech data, sidestepping the need for massive multimodal datasets.
