arXiv cs.AI / cs.LG / cs.CL·2d agoDo Audio Language Models Hear and Read Distinctive Features Alike?#audio-language-models#qwen2.5-omni#phonologyAI research