EVI
收藏资源简介:
EVI数据集是由位于英国伦敦的PolyAI Limited创建的多语言口语对话任务数据集,包含5,506个对话,涵盖英语、波兰语和法语。该数据集专注于基于知识的注册、验证和识别任务,旨在支持个性化和隐私导向的服务,通过口语对话系统进行用户身份的注册、验证和识别。数据集的创建过程涉及使用faker工具随机生成本地依赖的配置文件,并利用Google的特定语言语音识别和文本转语音技术。该数据集的应用领域包括自动化呼叫中心,以提高对话系统的安全性和用户体验。
The EVI dataset is a multilingual spoken dialogue task dataset developed by PolyAI Limited, headquartered in London, UK. It comprises 5,506 dialogues across English, Polish and French. This dataset centers on knowledge-based user identity registration, verification and recognition tasks, with the goal of supporting personalized and privacy-preserving services through spoken dialogue systems. The dataset's development process entails using the Faker tool to randomly generate locally tailored configuration files, and utilizing Google's language-specific speech recognition and text-to-speech technologies. Its application domains include automated call centers, where it helps enhance the security and user experience of dialogue systems.




