202小时德语游戏领域语音数据【数据堂】

德语游戏领域语音数据，选取了本国较为流行且长青的头部游戏中的对话语音，涵盖玩家商量对战策略、社交互动、电竞类新闻等内容，反映了真实世界的互动情境。此数据集标注了文本内容、冒犯性言论、说话人身份性别等多种属性，由多名来自不同地域和文化背景的人员录制，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，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

This is a German-language gaming domain speech dataset. It selects conversational speech from popular and long-standing top-tier games in the German market, covering content such as players discussing battle strategies, social interactions, esports news and other topics, which reflects real-world interactive scenarios. This dataset is annotated with multiple attributes including text transcripts, offensive speech labels, speaker identity and gender, etc. Recorded by multiple speakers from diverse regional and cultural backgrounds, it boasts high accuracy and strong usability, providing abundant resources for speech recognition-related research and applications. Verified by multiple AI companies, this dataset helps models perform excellently when facing the diversity of real-world scenarios. We strictly comply with data protection regulations and privacy rules, ensuring that user privacy and legitimate rights and interests are safeguarded during the collection, storage and use of the data. All data adheres to GDPR, CCPA and PIPL requirements.

搜集汇总

背景与挑战

背景概述

该数据集包含202小时的德语游戏领域语音数据，采集自德国流行游戏中的真实对话，覆盖策略讨论、社交互动等内容，并标注了文本、时间戳、冒犯性言论等多种属性。录音环境为混合场景，格式为16k Hz WAV单声道，词准确率达98%，专为语音识别研究设计，但仅限企业合作购买使用。