MetaSICL: Adapting Audiroty LLM via Meta Speech In-Context Learning

Auditory Large Language Models (LLMs) have demonstrated strong performance across a wide range of speech and audio understanding tasks. Nevertheless, they often struggle when applied to low-resource tasks. In case in-domain labeled data are scarce or mismatched with the true test distribution, direct fine-tuning can be brittle. In-Context Learning (ICL) provides a training-free, inference-time solution by adapting auditory LLMs through conditioning on a few in-domain demonstrations. In this work, we first show that $\mathit{\text{Vanilla}}\mathit{\text{ICL}}$, improves zero-shot performance across diverse speech and audio tasks for selected models which suggest that this ICL adaptation capability can be generalized to multimodal setting. Building on this, we propose $\mathbf{\text{Meta}}\mathbf{\text{Speech}}\mathbf{\text{In-Context}}\mathbf{\text{Learning}}\mathbf{\text{(MetaSICL)}}$, a post-training recipe utilizes only high resource speech data from various tasks intending to strengthen model’s in-context learning capability. Experiments indicate our proposed method outperforms direct fine-tuning in low-resource scenario.