Search papers, labs, and topics across Lattice.
1
0
3
6
Despite the intuition that noisy environments should push AVSR models to rely more on visual input, a persistent audio bias remains, even under severe degradation.
