Submitted by WangKe 22 VoiceAssistant-Eval: Benchmarking AI Assistants across Listening, Speaking, and Viewing LLMs for Reasoning 16 2