Paper accepted as an Oral at EMNLP 2025 (Short Paper, Main Track)! Visual-Aware Speech Recognition for Noisy Scenarios.