1 paper
Meng Chen, Kun Wang, Li Lu +2
Modern Large audio-language models (LALMs) power intelligent voice interactions by tightly integrating audio and text. This integration, however, expands the attack surface beyond…