cdli/common_voice_swahili_small

这是一个从Common Voice 20.0派生出来的小型数据集，通过筛选具有最小点赞数的语音样本。数据集包括说话者ID、音频ID、转录文本、句子领域、点赞数、点踩数、年龄、性别、口音、变体、地区、片段时长和音频文件等信息。数据集分为训练集、测试集和验证集，分别包含4548、397和335个样本，遵循Creative Commons Zero许可。

This is a small dataset derived from Common Voice 20.0 by selecting utterances with minimum upvote counts. The dataset includes speaker ID, audio ID, transcription, sentence domain, up votes, down votes, age, gender, accents, variant, locale, segment duration, and audio file with its sampling rate and length. The dataset is split into training, test, and validation sets with 4548, 397, and 335 samples respectively, and is released under the Creative Commons Zero licence.