3,881小时普通话语音数据_口语化【数据堂】

普通话语音数据_口语化，内容覆盖体育、对话、综艺等领域。反映了真实世界的互动情境。此数据集标注了文本内容、说话人身份性别等多种属性，由来自不同地域和文化背景的中国本土人录制，准确性高，易用性强，为语音识别相关研究及应用提供了丰富的资源，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

Colloquial Mandarin speech dataset covering domains including sports, dialogues, and variety shows. It reflects real-world interactive scenarios. This dataset is annotated with multiple attributes such as transcript content, speaker identity and gender. Recorded by native Chinese speakers from different regions and cultural backgrounds, it features high accuracy and strong usability, providing rich resources for speech recognition-related research and applications, and enabling models to perform excellently when confronted with the diversity of real-world scenarios. We strictly abide by data protection laws and privacy regulations, ensuring that user privacy and legitimate rights and interests are protected throughout the entire process of data collection, storage and usage. All data complies with GDPR, CCPA and PIPL.

搜集汇总

背景与挑战

背景概述

该数据集包含3,881小时普通话口语化语音数据，覆盖体育、对话、综艺等领域，由不同背景的中国本土人在低噪环境中录制，格式为16kHz、16 bit、wav单声道，适用于语音识别研究。数据标注了文本内容、句时间戳、说话人标识、性别和噪音，句准确率达95%，但仅作为商业数据对企业合作开放。