Audio source separation datasets
收藏DataCite Commons2024-08-16 更新2025-04-16 收录
下载链接：
https://ieee-dataport.org/documents/audio-source-separation-datasets
下载链接
链接失效反馈官方服务：
资源简介：
In the experiments, audio signals from three public datasets are used: Librispeech (Dataset A), MSD100 (Dataset B), and MS-SNSD (Dataset C). Dataset A consists of 1000 hours of read English speech from the LibriVox project, sampled at 16 kHz. Dataset B, MSD100, includes 100 full-track songs with various vocals and accompaniments, resampled to 16 kHz. Dataset C, MS-SNSD, contains various noises recorded at 16 kHz. Both synthetic and authentic room impulse responses (RIRs) are employed to simulate different reverberation scenarios, with the synthetic RIRs set in a virtual room and the authentic RIRs measured in a controlled acoustic lab.
提供机构：
IEEE DataPort
创建时间：
2024-08-16
