arXiv 2412.207222025-04-28eess.AScs.SD

低资源条件下的声学场景分类改进

Improving Acoustic Scene Classification in Low-Resource Conditions

Jiangsu Normal University（江苏师范大学）；Department of Electronic Engineering, Tsinghua University（清华大学电子工程系）

Zhi Chen, Yun-Fei Shao, Yong Ma, Mingsheng Wei, Le Zhang, Wei-Qiang Zhang

备注：: Copyright 2025 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component
发表信息：: ICASSP (2025)
DOI：: 10.1109/ICASSP49660.2025.10888928

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

本文提出DS-FlexiNet模型，通过结合深度可分离卷积与残差连接，并采用量化感知训练、数据增强及知识蒸馏技术，有效提升了低资源及跨设备条件下声学场景分类的效率、适应性与准确率。

摘要：

声学场景分类（ASC）基于音频信号识别环境。本文探索低资源条件下的ASC，提出新模型DS-FlexiNet，结合MobileNetV2的深度可分离卷积与受ResNet启发的残差连接，以平衡效率与精度。为解决硬件限制和设备异构性，DS-FlexiNet采用量化感知训练（QAT）进行模型压缩，并使用自动设备脉冲响应（ADIR）和Freq-MixStyle（FMS）等数据增强方法提升跨设备泛化能力。来自12个教师模型的知识蒸馏（KD）进一步增强了在未见设备上的性能。该架构包含自定义残差归一化层以处理跨设备域差异，深度可分离卷积在不牺牲特征表示的情况下降低计算开销。实验结果表明，DS-FlexiNet在资源受限条件下的适应性和性能均表现优异。

英文摘要：

Acoustic Scene Classification (ASC) identifies an environment based on an audio signal. This paper explores ASC in low-resource conditions and proposes a novel model, DS-FlexiNet, which combines depthwise separable convolutions from MobileNetV2 with ResNet-inspired residual connections for a balance of efficiency and accuracy. To address hardware limitations and device heterogeneity, DS-FlexiNet employs Quantization Aware Training (QAT) for model compression and data augmentation methods like Auto Device Impulse Response (ADIR) and Freq-MixStyle (FMS) to improve cross-device generalization. Knowledge Distillation (KD) from twelve teacher models further enhances performance on unseen devices. The architecture includes a custom Residual Normalization layer to handle domain differences across devices, and depthwise separable convolutions reduce computational overhead without sacrificing feature representation. Experimental results show that DS-FlexiNet excels in both adaptability and performance under resource-constrained conditions.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。