AI RESEARCH
Towards Holistic Evaluation of Large Audio-Language Models: A Comprehensive Survey
arXiv CS.AI
•
ArXi:2505.15957v4 Announce Type: replace-cross With advancements in large audio-language models (LALMs), which enhance large language models (LLMs) with auditory capabilities, these models are expected to nstrate universal proficiency across various auditory tasks. While numerous benchmarks have emerged to assess LALMs' performance, they remain fragmented and lack a structured taxonomy.