DCASE-18
收藏arXiv2025-09-30 收录
下载链接：
https://github.com/AlbertoAncilotto/NeSsi
下载链接
链接失效反馈官方服务：
资源简介：
该数据集是2018年TUT城市声学场景的开发训练集，用于音频场景分类任务。每个音频录音的时长为10秒，输入数据为大小为96×64的对数梅尔频谱图。该训练集的规模旨在对10种不同的音频场景进行分类。
This dataset is the development training set of the TUT Urban Acoustic Scenes 2018 corpus, which is utilized for audio scene classification tasks. Each audio recording has a duration of 10 seconds, and the input data consists of 96×64 log-mel spectrograms. This training set is designed for classifying 10 distinct categories of audio scenes.
提供机构：
TUT
