GTSinger
收藏资源简介:
GTSinger是由浙江大学创建的一个大型全球多技术歌唱语料库,旨在支持所有歌唱任务。该数据集包含80.59小时的优质歌唱语音,由20位专业歌手使用六种常见歌唱技术录制,涵盖九种广泛使用的语言。数据集的创建过程包括录音、对齐、技术与风格注释以及真实音乐乐谱的编写。GTSinger不仅提供了高质量的歌唱语音,还包含了详细的语音对齐、全局风格标签和16.16小时的配对语音,适用于多种歌唱任务,如技术可控的歌唱语音合成、技术识别、风格转换和语音到歌唱的转换。
GTSinger is a large-scale global multi-technical singing corpus developed by Zhejiang University, aiming to support all singing-related tasks. This corpus contains 80.59 hours of high-quality singing voices, recorded by 20 professional singers using six common singing techniques, and covers nine widely used languages. The construction process of the dataset includes recording, alignment, technical and style annotation, as well as the compilation of authentic musical scores. In addition to providing high-quality singing voices, GTSinger also includes detailed voice alignment, global style tags, and 16.16 hours of paired voice data, which is applicable to various singing tasks such as technique-controlled singing voice synthesis, technique recognition, style transfer, and voice-to-singing conversion.
- 1GTSinger: A Global Multi-Technique Singing Corpus with Realistic Music Scores for All Singing Tasks浙江大学 · 2024年



