SEE: Signal Embedding Energy for Quantifying Noise Interference in Large Audio Language Models

2026-01-12

AI Summary

This work addresses the limited robustness of large audio language models (LALMs) in noisy environments and the absence of effective quantitative evaluation metrics. The authors propose Signal Embedding Energy (SEE), a novel method that quantifies the perturbation of input representations under noise by analyzing structured subspaces within LALM internal activations. SEE demonstrates a strong correlation (Pearson’s r = 0.98) with model performance, revealing a critical mismatch between conventional speech denoising techniques and the noise sensitivity characteristics of LALMs. Leveraging this insight, the authors design a targeted denoising strategy tailored to LALM-specific representation dynamics. Experimental results show that this approach significantly outperforms existing methods, substantially enhancing LALM robustness in real-world noisy conditions.

Technology Category

• Natural Language Processing: Safety and Robustness
• Machine Learning: Large Multimodal Models (LMMs)
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Large language models for search
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Security and Privacy: Large-scale security measurements

Abstract

Large Audio Language Models (LALMs) have been widely applied in real-time scenarios, such as in-car assistants and online meeting comprehension. In practice, audio inputs are often corrupted by device and environmental noise, leading to performance degradation. However, existing LALM studies on noise lack quantitative analysis and rely mainly on intuition and empirical observation, thus failing to understand practical robustness. To address this issue, we introduce Signal Embedding Energy (SEE), a method for quantifying the impact of noise intensity on LALM inputs, enabling the differentiation of LALM robustness in real-world deployments. SEE introduces a perspective based on structured activation subspaces derived from the model's internal representations, which more accurately captures its perception of noise than raw audio features. Across experiments, SEE exhibits a strong correlation with LALM performance, achieving a correlation of 0.98. Surprisingly, traditional audio denoising methods are only marginally effective for LALMs, and, in some cases, even increase SEE and impair performance. This suggests a mismatch between speech-centric denoising objectives and the noise sensitivity of modern LALMs. Therefore, we propose a mitigation strategy derived from SEE to denoise LALM inputs, outperforming existing denoising methods. This paper introduces a novel metric for noise quantification in LALMs, providing guidance for robustness improvements in real-world deployments.

Problem

• Large Audio Language Models
• noise interference
• robustness
• quantitative analysis
• Signal Embedding Energy

Innovation

• Signal Embedding Energy
• Large Audio Language Models
• Noise Robustness
• Activation Subspace
• Denoising Strategy