pklumpp/CommonPhoneDataset

Common Phone是一个多语言语音数据集，包含11650小时的语音样本，由11146位说话者在六种语言中收集而成。该数据集旨在为机器学习研究人员和爱好者提供一个基础，以训练和测试他们的模型，以改进机器学习在现实世界语音应用中的泛化能力和可用性。数据集支持六种不同的语言，提供时间对齐的国际音标(IPA)音位转录，以及性别平衡和年龄段匹配的说话者分布。

Common Phone is a multilingual speech dataset comprising 11650 hours of speech samples collected from 11146 speakers across six languages. This corpus is designed to provide a foundation for ML researchers and enthusiasts to train and test their models, improving the generalization and availability of ML in real-world speech applications. The dataset supports six different languages, offers time-aligned International Phonetic Alphabet (IPA) phonetic transcriptions, and features a gender-balanced and age-group-matched distribution of speakers.