道德语音识别数据集的注意事项（英文版）

英文标题：Considerations for Ethical Speech Recognition Datasets中文摘要：本文以自动语音识别为例，研究具备道德操守的语音数据集应具备的特征，包括多样性、包容性、关键元数据等，以提高模型鲁棒性与可用性，同时保护用户隐私、提升可解释性、促进社会技术与情境模型设计，旨在鼓励研究者与实践者建立更多人性化的数据集，提高模型效果。英文摘要：Speech AI Technologies are largely trained on publicly available datasets orby the massive web-crawling of speech. In both cases, data acquisition focuseson minimizing collection effort, without necessarily taking the data subjects'protection or user needs into consideration. This results to models that are