JackyHoCL/common_voice_22_yue_w_background_caption

该数据集是一个音频数据集，包含多个分割（训练、验证、测试、已验证、未验证和其他）。数据集的特征包括音频（采样率16000）、客户端ID、路径、句子、句子ID、句子领域、投票数（赞成和反对）、年龄、性别、口音、变体、地区、片段、标题和背景标题。数据集合并了JackyHoCL/urban-noise-uganda-61k-caption和OpenSound/AudioCaps两个数据集，并计划转换为MP3格式以减少大小。

This dataset is an audio dataset containing multiple splits (train, validation, test, validated, invalidated, and other). The features of the dataset include audio (sampling rate 16000), client ID, path, sentence, sentence ID, sentence domain, votes (up and down), age, gender, accents, variant, locale, segment, caption, and background caption. The dataset merges JackyHoCL/urban-noise-uganda-61k-caption and OpenSound/AudioCaps, with plans to convert to MP3 format to reduce size.