EARS (Expressive Anechoic Recordings of Speech)

EARS数据集是由汉堡大学信号处理团队创建的高质量语音数据集，包含107位来自不同背景的说话者，总计100小时的清晰无回声语音数据。数据集涵盖了广泛的说话风格，包括情感语音、不同阅读风格、非言语声音和自由形式的对话语音。创建过程中，语音在无回声室中以48kHz的采样率录制，确保了高录音质量。该数据集主要用于语音增强和去混响的研究，旨在解决语音处理中的清晰度和质量问题。

The EARS dataset is a high-quality speech dataset created by the Signal Processing Team at the University of Hamburg. It includes 100 hours of clear, anechoic speech data from 107 speakers with diverse backgrounds. The dataset covers a wide range of speaking styles, including emotional speech, various reading styles, non-verbal vocalizations, and free-form conversational speech. During the data collection process, the speech was recorded in an anechoic chamber at a sampling rate of 48 kHz to ensure high recording quality. This dataset is primarily used for research on speech enhancement and dereverberation, aiming to address issues of speech clarity and quality in speech processing.