@misc{arxiv_2610_25352v1,
  title = {Local Evidence, Global Failure: Diagnosing Speech-Vision Temporal Composition in Audio-Visual Language Models},
  author = {{Jaehong Myung} and {Seunghun Oh} and {Unsang Park}},
  eprint = {2610.25352v1},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2610.25352v1}
}