ÌròyìnSpeech

ÌròyìnSpeech是一个旨在增加高质量当代约鲁巴语语音数据量的新型语料库，适用于文本到语音（TTS）和自动语音识别（ASR）任务。该数据集由斯坦福大学等机构创建，包含约23000条从新闻和创意写作领域精选的文本句子，以及42小时的语音数据，由80名志愿者录制。数据集创建过程中，采用了参与式方法，将5000条句子提供给Mozilla Common Voice平台进行众包录音和验证。ÌròyìnSpeech的应用领域主要集中在提升约鲁巴语语音技术的质量和可用性，解决非洲语言在语音技术领域的代表性不足问题。ÌròyìnSpeech is a novel corpus designed to increase the scale of high-quality contemporary Yoruba speech data, suitable for text-to-speech (TTS) and automatic speech recognition (ASR) tasks. Developed by institutions including Stanford University, this dataset contains approximately 23,000 textual sentences selected from news and creative writing domains, as well as 42 hours of speech data recorded by 80 volunteers. During the dataset construction, a participatory approach was adopted, where 5,000 sentences were provided to the Mozilla Common Voice platform for crowdsourced recording and validation. The core application scenarios of ÌròyìnSpeech focus on improving the quality and availability of Yoruba speech technologies, addressing the underrepresentation issue of African languages in the field of speech technology.