CHiME-5

CHiME-5数据集由约翰霍普金斯大学语言与语音处理中心创建，专注于在真实家庭环境中进行远场多麦克风对话自动语音识别（ASR）的研究。该数据集包含20个不同家庭中的晚餐派对录音，总计79,980条语音数据，使用6个Kinect麦克风阵列和4对双耳麦克风进行录制。数据集的创建过程涉及在厨房、餐厅和客厅三个不同场景下自然对话的捕捉。CHiME-5数据集主要用于推动ASR技术在噪声和混响环境下的发展，特别是在家庭自动化和多媒体系统中的应用。

The CHiME-5 dataset was developed by the Center for Language and Speech Processing at Johns Hopkins University, focusing on research into far-field multi-microphone conversational automatic speech recognition (ASR) in real-world home environments. It encompasses dinner party recordings from 20 unique households, totaling 79,980 speech utterances, and was captured using 6 Kinect microphone arrays and 4 pairs of binaural microphones. The dataset creation process involves collecting natural conversations across three distinct scenarios: the kitchen, dining room, and living room. The CHiME-5 dataset is primarily used to advance the development of ASR technologies in noisy and reverberant environments, especially for applications in home automation and multimedia systems.