VoxPopuli ASR dataset

该数据集名为VoxPopuli ASR，主要用于多说话人的语音识别，并生成训练语音正常化的参考单元。该数据集结合文本转录，创建用于参考目标单元。它包含了4500小时英语、西班牙语和法语的数据，总计13500小时，适用于自动语音识别任务。

The dataset named VoxPopuli ASR is primarily intended for multi-speaker automatic speech recognition, and is used to generate reference units for speech normalization training. It incorporates text transcriptions to create reference target units, and contains 4,500 hours of data per language for English, Spanish and French, totaling 13,500 hours. This dataset is applicable to automatic speech recognition tasks.