kosp2e
收藏资源简介:
kosp2e是一个专为韩国语音到英语文本翻译设计的语料库,由首尔国立大学电气与计算机工程系及INMC创建。该数据集包含约110,239条翻译后的语音数据,涵盖新闻、教科书、AI代理命令和日记等多个领域。数据集的创建过程涉及对公开许可的语音识别、机器翻译和口语语料库的采用,并通过社区贡献进行补充。kosp2e的应用领域广泛,包括国际会议、视频字幕和实时翻译等,旨在解决非英语母语者在语音翻译技术中的使用障碍。
Kosp2e is a corpus specifically designed for Korean speech-to-English text translation, developed by the Department of Electrical and Computer Engineering of Seoul National University and INMC. This dataset contains approximately 110,239 translated speech data entries, covering multiple domains including news, textbooks, AI agent commands, and diaries. The construction of the dataset utilizes openly licensed speech recognition, machine translation, and spoken language corpora, and is supplemented by community contributions. Kosp2e has a wide range of application scenarios, including international conferences, video subtitling, and real-time translation, aiming to address the usage barriers faced by non-native English speakers in speech translation technologies.

- 1Kosp2e: Korean Speech to English Translation Corpus首尔国立大学电气与计算机工程系及INMC · 2021年



