LRWR
收藏arXiv2021-09-14 更新2024-08-06 收录
下载链接：
http://arxiv.org/abs/2109.06692v1
下载链接
链接失效反馈官方服务：
资源简介：
LRWR是由莫斯科物理技术学院创建的俄语唇读数据集，包含235个类别和135位说话者，总计超过117500个样本。该数据集通过从YouTube视频中收集，涵盖了广泛的说话者环境和话题，如历史、艺术、旅行等。数据集的创建过程包括视频筛选、词汇准备和数据预处理，确保了数据的多样性和真实性。LRWR旨在解决俄语环境下的自动语音识别问题，特别是在嘈杂环境或多说话者场景中的应用。
LRWR is a Russian lip-reading dataset created by the Moscow Institute of Physics and Technology. It includes 235 categories and 135 speakers, with a total of over 117,500 samples. Collected from YouTube videos, the dataset covers a wide range of speaker contexts and topics such as history, art, travel and others. The dataset's construction process involves video screening, vocabulary preparation and data preprocessing, which ensures the diversity and authenticity of the data. LRWR aims to solve automatic speech recognition problems in Russian contexts, especially for applications in noisy environments or multi-speaker scenarios.
提供机构：
莫斯科物理技术学院
创建时间：
2021-09-14
