672小时东北话自然对话（双语标注）手机采集语音数据

东北话自然对话（双语标注）手机采集语音数据，基于30余个常见主题来模拟录制。此数据集标注了方言及普通话两版文本内容、句时间戳、说话人身份、性别等多种属性，由来自东北地区的方言使用者录制，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。Northeastern Mandarin Natural Dialogue (Bilingual Annotated) Mobile-collected Speech Dataset is developed by simulating natural dialogues based on over 30 common daily topics. This dataset is annotated with comprehensive attributes including bilingual transcriptions (Northeastern Mandarin and Standard Mandarin), sentence-level timestamps, speaker identity, gender and other relevant information. It is recorded by native Northeastern Mandarin speakers from Northeast China, ensuring high data accuracy. This dataset offers rich resources for research and applications related to speech recognition, and has been validated by multiple AI companies that it can help models achieve excellent performance when dealing with real-world diversity. We strictly comply with global and regional data protection regulations and privacy rules, ensuring the protection of users' privacy and legitimate rights and interests throughout the entire lifecycle of data collection, storage and usage. All data in this dataset adheres 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672小时的东北话自然对话语音数据，通过手机采集，标注了方言和普通话双语文本、句时间戳及说话人信息，覆盖30余个常见主题，由808名东北地区说话人录制，男女比例均衡，准确率高，适用于语音识别和声纹识别研究。数据集具有大规模、多主题、真实对话的特点，并遵循严格的数据隐私法规，为AI模型训练提供了丰富的方言资源。