Adaptive Discovery of Interpretable Audio Attributes with Multimodal LLMs for Low-Resource Classification
This work addresses the bottleneck of low-throughput human-driven attribute discovery for low-resource audio classification, which is crucial for high-reliability applications requiring interpretable attributes.
This paper proposes a method using Multimodal Large Language Models (MLLMs) to adaptively discover interpretable audio attributes for low-resource audio classification. By replacing human-driven attribute discovery with MLLMs, the method achieves significantly faster attribute discovery, completing training within 11 minutes, and outperforms direct MLLM prediction in most evaluated cases.
In predictive modeling for low-resource audio classification, extracting high-accuracy and interpretable attributes is critical. Particularly in high-reliability applications, interpretable audio attributes are indispensable. While human-driven attribute discovery is effective, its low throughput becomes a bottleneck. We propose a method for adaptively discovering interpretable audio attributes using Multimodal Large Language Models (MLLMs). By replacing humans in the AdaFlock framework with MLLMs, our method achieves significantly faster attribute discovery. Our method dynamically identifies salient acoustic characteristics via prompting and constructs an attribute-based ensemble classifier. Experimental results across various audio tasks demonstrate that our method outperforms direct MLLM prediction in the majority of evaluated cases. The entire training completes within 11 minutes, proving it a practical, adaptive solution that surpasses conventional human-reliant approaches.