Revolab/AudioSet-Chunk
收藏Hugging Face2025-03-27 更新2025-04-08 收录
下载链接：
https://hf-mirror.com/datasets/Revolab/AudioSet-Chunk
下载链接
链接失效反馈官方服务：
资源简介：
AudioSet-Chunk数据集包含了每0.25秒和0.5秒分割的AudioSet音频片段，用于音频分类任务。数据集由四个部分组成：chunk_0.5s、500k_part1_chunk_0.5s、chunk_0.25s和500k_part2_chunk_0.5s，每个部分包含了大量的音频样本。使用MIT/ast-finetuned-audioset-10-10-0.4593模型对音频片段进行预测，并筛选出与黄金标签重合的标签。
The AudioSet-Chunk dataset consists of AudioSet audio clips segmented every 0.25 seconds and 0.5 seconds for audio classification tasks. The dataset is divided into four parts: chunk_0.5s, 500k_part1_chunk_0.5s, chunk_0.25s, and 500k_part2_chunk_0.5s, each containing a large number of audio samples. The MIT/ast-finetuned-audioset-10-10-0.4593 model is used to predict the audio clips, and labels overlapping with the gold labels are filtered out.
提供机构：
Revolab
