ROBIN Technical Acquisition Speech Corpus (ROBINTASC)

ROBIN Technical Acquisition Speech Corpus (ROBINTASC) 是由罗马尼亚科学院人工智能研究所创建的一个新的罗马尼亚语语音数据集，旨在改善对话代理的行为，实现购买技术设备时的人机交互。该数据集包含6.5小时的录音，由6名不同性别和年龄的演讲者完成，总计3786个音频文件。数据集的创建过程包括使用RELATE平台进行录音和文本上传，以及使用UDPipe进行文本标注。ROBINTASC主要应用于ROBIN项目中的自动语音识别系统和对话管理器，以解决在电子商店笔记本部门的人机交互问题。

ROBIN Technical Acquisition Speech Corpus (ROBINTASC) is a novel Romanian speech dataset developed by the Artificial Intelligence Institute of the Romanian Academy. It aims to optimize the performance of conversational agents and enable natural human-computer interaction during the purchase of technical equipment. The dataset contains 6.5 hours of recorded audio, completed by 6 speakers with distinct genders and age groups, totaling 3786 audio files. The creation of ROBINTASC involved utilizing the RELATE platform for recording and text upload, as well as UDPipe for text annotation. This corpus is primarily applied to automatic speech recognition systems and dialogue managers in the ROBIN project, to address human-computer interaction issues in the laptop department of electronics retail stores.