ggfox00000/stt-voxpopuli-test-en

VoxPopuli EN测试分割是一个用于自动语音识别（ASR）任务的英语数据集，包含1842个欧洲议会（2009-2020年）的语音片段。音频为16kHz单声道WAV格式，并附有原始文本和标准化文本的转录。标准化文本是首选的参考转录文本。数据集还包括说话者的性别、ID、口音信息以及是否为人工验证的黄金转录。该数据集来源于Wang, Riviere等人在ACL 2021发表的论文。VoxPopuli EN — test split is an English dataset for automatic speech recognition (ASR) tasks, containing 1842 speech segments from the European Parliament (2009-2020). The audio is in 16kHz mono WAV format, accompanied by raw and normalized text transcriptions, with the normalized text being the preferred reference. The dataset also includes speaker gender, ID, accent information, and whether the transcription is human-validated (gold transcript). The dataset originates from the paper by Wang, Riviere et al. at ACL 2021.