Zambezi Voice

Zambezi Voice是一个为赞比亚语言开发的开源多语言语音数据集，由赞比亚大学创建。该数据集包含两个部分：未标记的音频记录（160小时）和标记数据（超过80小时），用于语音识别研究。数据集涵盖了Bemba、Nyanja、Tonga和Lozi等四种语言，旨在支持低资源语言的语音处理研究。创建过程中，数据来源于公开可用的文献书籍和广播节目，通过LIG-AIKUMA移动应用程序进行记录。该数据集的应用领域包括语音识别和多语言语音处理，特别是针对资源匮乏的语言环境。

Zambezi Voice is an open-source multilingual speech dataset developed by the University of Zambia for Zambian languages. This dataset comprises two subsets: 160 hours of unlabeled audio recordings and over 80 hours of labeled data, dedicated to speech recognition research. Covering four languages including Bemba, Nyanja, Tonga and Lozi, it aims to support speech processing research for low-resource languages. During its creation, the data was sourced from publicly available literature books and broadcast programs, and recorded via the LIG-AIKUMA mobile application. Its application domains include speech recognition and multilingual speech processing, particularly in resource-constrained linguistic environments.