Serial Speakers
收藏arXiv2020-02-17 更新2024-08-06 收录
下载链接：
http://arxiv.org/abs/2002.06923v1
下载链接
链接失效反馈官方服务：
资源简介：
Serial Speakers数据集由阿维尼翁大学信息实验室创建，包含161集来自三部美国热门电视剧的标注数据，包括《绝命毒师》、《权力的游戏》和《纸牌屋》。该数据集适用于多媒体检索和语音处理研究，提供每段对话的边界、说话者和场景边界等详细标注。由于版权限制，数据集中的文本内容被加密，但提供了一个在线工具，允许用户使用自己的字幕文件恢复原始文本。数据集旨在解决电视剧连续性叙事分析的挑战，支持多媒体检索和语音处理任务。
Serial Speakers Dataset was developed by the Information Laboratory of Avignon University. It contains annotated data from 161 episodes of three popular American television series: *Breaking Bad*, *Game of Thrones*, and *House of Cards*. Targeted at multimedia retrieval and speech processing research, this dataset provides detailed annotations including dialogue boundaries, speaker identities, and scene boundaries for each conversation segment. Due to copyright restrictions, the textual content within the dataset is encrypted, and an online tool is offered to enable users to restore the original text using their own subtitle files. This dataset is designed to address the challenges in continuous narrative analysis of TV series, and supports multimedia retrieval and speech processing tasks.
提供机构：
阿维尼翁大学信息实验室
创建时间：
2020-02-17
