AudioCaps

该数据集名为AudioCaps，包含了大约47,000个长度为十秒的音频数据，这些数据涵盖了更加多样化的声音事件。此外，该数据集被包含在AudioLDM的预训练中，其规模属于大型，任务类型为文本到声音的生成。

This dataset is named AudioCaps. It comprises approximately 47,000 ten-second audio clips that cover a more diverse range of sound events. Additionally, this large-scale dataset is included in the pre-training process of AudioLDM, and its target task is text-to-audio generation.

搜集汇总

背景与挑战

背景概述

AudioCaps是一个大规模音频字幕生成数据集，包含约46,000个音频片段与人工撰写的文本配对，用于探索音频表示和字幕生成模型的有效性。