Scholarlink

Acoustic scene classification (ASC) is a critical task for understanding the environmental sound applications such as smart home, monitoring process, and sound detection. Conventional ASC systems cannot fit for diverse real-world audio recordings because the systems depend on default 10-s segments for analysis. In this paper, the granularity analysis is performed with varied audio segment durations, ranging from 1 to 10 s, to capture the diverse acoustic scenes thoroughly. In addition, deep residual learning network with channel attention mechanisms (CA_DRLN) is applied to increase the performance of ASC system. With systematic analysis, the proposed ASC system can provide high-performance results compared to the traditional techniques with detailed granularity analysis. This study uncovers the advanced ASC approach, highlighting the impacts of audio analysis on ASC domain.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhancing Acoustic Scene Classification with Channel Attention Based on Deep Residual Learning Network

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Mie Mie Oo,
  • Nu War

摘要

Acoustic scene classification (ASC) is a critical task for understanding the environmental sound applications such as smart home, monitoring process, and sound detection. Conventional ASC systems cannot fit for diverse real-world audio recordings because the systems depend on default 10-s segments for analysis. In this paper, the granularity analysis is performed with varied audio segment durations, ranging from 1 to 10 s, to capture the diverse acoustic scenes thoroughly. In addition, deep residual learning network with channel attention mechanisms (CA_DRLN) is applied to increase the performance of ASC system. With systematic analysis, the proposed ASC system can provide high-performance results compared to the traditional techniques with detailed granularity analysis. This study uncovers the advanced ASC approach, highlighting the impacts of audio analysis on ASC domain.