Investigating Safety Vulnerabilities of Large Audio-Language Models Under Speaker Emotional Variations Paper • 2510.16893 • Published Oct 19, 2025 • 17