tii-audio/masc-benchmark

这是一个包含视频信息的语音识别数据集，其中包括视频ID、起始时间、结束时间、持续时间、对话信息（包括对话者from和对话内容value）、音频文件路径、地面真实值标签和版本号。数据集分为两个部分：clean_test和noisy_test，分别用于测试在干净和嘈杂环境下的语音识别性能。

This is a speech recognition dataset containing video information, including video ID, start time, end time, duration, conversation information (including speaker and content), audio file path, ground truth label, and version number. The dataset is divided into two parts: clean_test and noisy_test, for testing speech recognition performance in clean and noisy environments respectively.