量化音频模型在零样本伪造检测中表现不佳，但FP16可兼顾效率与精度。
Can Quantized Audio Language Models Perform Zero-Shot Spoofing Detection?
- 测试五种量化音频模型在三数据集上的零样本检测能力。
- 所有模型存在严重伪造分类偏差，实际性能接近随机判断。
- FP16量化几乎不降精度，内存和计算减半，适合部署。
量化对在资源受限环境中高效部署大型音频语言模型（LALMs）至关重要，但其对复杂任务如零样本音频伪造检测的影响仍不明确。本研究评估了GAMA、LTU-AS、MERaLiON、Qwen-Audio和SALMONN五种LALMs在ASVspoof2019、In-the-Wild和WaveFake三个数据集上的零样本能力，并探究其对量化（FP32、FP16、INT8）的鲁棒性。尽管初始检测准确率较高，但所有模型均表现出严重的伪造分类偏差，导致实际性能等同于随机分类。有趣的是，将模型量化至FP16精度时性能下降可忽略，有效降低一半内存与计算需求，且不影响准确率；而INT8量化加剧了模型偏差，显著降低平衡准确率。结果揭示了关键的架构局限性，并强调FP16量化是实用部署的理想折衷方案，为未来模型优化提供指导。
原文摘要 · Abstract (English)
Quantization is essential for deploying large audio language models (LALMs) efficiently in resource-constrained environments. However, its impact on complex tasks, such as zero-shot audio spoofing detection, remains underexplored. This study evaluates the zero-shot capabilities of five LALMs, GAMA, LTU-AS, MERaLiON, Qwen-Audio, and SALMONN, across three distinct datasets: ASVspoof2019, In-the-Wild, and WaveFake, and investigates their robustness to quantization (FP32, FP16, INT8). Despite high initial spoof detection accuracy, our analysis demonstrates severe predictive biases toward spoof classification across all models, rendering their practical performance equivalent to random classification. Interestingly, quantization to FP16 precision resulted in negligible performance degradation compared to FP32, effectively halving memory and computational requirements without materially impacting accuracy. However, INT8 quantization intensified model biases, significantly degrading balanced accuracy. These findings highlight critical architectural limitations and emphasize FP16 quantization as an optimal trade-off, providing guidelines for practical deployment and future model refinement.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。