数据堂—275小时韩英混读语音数据_朗读（手机）

韩英混读语音数据_朗读（手机），基于给定的脚本朗读并模拟录制，内容是韩语和英文混杂的句子。此数据集标注了文本内容，由737名韩国本土人使用苹果、安卓手机录制，准确性高，为语音识别相关研究及应用提供了丰富的资源，经多家AI公司验证：有助于模型面对真实世界的多样性时能够表现出色。我们严格遵循数据保护法规和隐私规定，确保数据采集、存储和使用的过程中维护用户的隐私和合法权益，所有数据均遵循GDPR, CCPA, PIPL

Korean-English Code-Switching Read Speech Dataset (Mobile-Recorded)
This dataset comprises speech data collected by 737 native Korean speakers, who read aloud pre-defined scripts containing mixed Korean and English sentences via Apple and Android mobile devices. Each speech sample is annotated with its corresponding text transcript. Featuring high recording accuracy, this dataset serves as a valuable resource for research and applications in the field of automatic speech recognition (ASR). Verified by multiple AI enterprises, it enables ASR models to achieve robust performance when encountering real-world linguistic diversities. We strictly comply with global data protection and privacy regulations, safeguarding users' privacy and legitimate rights throughout the entire lifecycle of data collection, storage and utilization. All dataset records adhere to the requirements of GDPR, CCPA and PIPL.

搜集汇总

背景与挑战

背景概述

该数据集包含275小时的韩英混合朗读语音，由737名韩国母语者使用智能手机在安静室内录制，适用于语音识别研究与应用。数据以16kHz WAV格式提供，包含文本转录，句子级准确率达95%，并遵循GDPR等数据保护法规。