Search papers, labs, and topics across Lattice.
University of Iowa Iowa
1
0
2
Task-specific audio perturbations can boost model accuracy by over 6% and reduce hallucinations in large audio-language models.
