RyanSpeech

RyanSpeech是一个专为对话式文本到语音合成研究设计的高质量男性语音数据集。该数据集由丹佛大学电气与计算机工程系和DreamFace科技有限公司共同创建，包含超过10小时的44.1 kHz采样率的录音，由专业男性声优录制。数据集内容涵盖多种真实对话场景，如电影、体育、音乐等，旨在解决现有公开TTS数据集中的噪音和多说话者问题。RyanSpeech的创建过程包括从多个文本资源中收集材料，进行文本预处理和音频后处理，确保数据集的质量和适用性。该数据集特别适用于开发高质量的TTS系统，尤其是在需要自然语音合成的应用领域，如电影和播客制作。

RyanSpeech is a high-quality male speech dataset specifically designed for conversational text-to-speech synthesis research. Co-developed by the Department of Electrical and Computer Engineering, University of Denver and DreamFace Technology Co., Ltd., it contains over 10 hours of recordings with a 44.1 kHz sampling rate, captured by professional male voice actors. The dataset covers a wide range of real conversational scenarios including movies, sports, music and other domains, aiming to resolve the noise and multi-speaker problems prevalent in current public TTS datasets. The development workflow of RyanSpeech involves collecting materials from multiple text resources, conducting text preprocessing and audio post-processing to guarantee the dataset's quality and applicability. This dataset is particularly well-suited for developing high-quality TTS systems, especially in application fields requiring natural speech synthesis such as movie and podcast production.