TUT Urban Acoustic Scenes 2018 development dataset (DCASE-18)
收藏arXiv2025-09-30 收录
下载链接:
https://dcase.community/challenge2018/task-acoustic-scene-classification
下载链接
链接失效反馈官方服务:
资源简介:
该数据集是为了对10种不同的声学场景进行分类而设计的。输入到VGGish_Net的是一个大小为(96×64)的对数梅尔频谱图,该频谱图对应于整个10秒声学场景中的一个960毫秒音频片段。该数据集是公开可用的,其任务是对声学场景进行分类。
This dataset is designed for the classification task of 10 distinct acoustic scenes. The input to VGGish_Net is a log-Mel spectrogram with dimensions of (96 × 64), which corresponds to a 960-millisecond audio segment extracted from the full 10-second acoustic scene. This dataset is publicly available, and its core task is acoustic scene classification.
提供机构:
DCASE 2018



