DAMP-VSEP: Smule Digital Archive of Mobile Performances - Vocal Separation

This dataset contains audio content for performing vocal separation research. It contains performances with (1) backing tracks, (2) one or more isolated vocals files, and (3) a mixture of the two. This is a real-world dataset from performances on the Smule app, which includes non-linear effects processes on the mixture of vocals and backing track. The dataset contains data from 155 countries, 36 languages, 6456 artists, and 11494 compositions. Users of this dataset must read and accept Smule's Research Data License Agreement (LICENSE.txt).

搜集汇总

背景与挑战

背景概述

DAMP-VSEP是一个用于语音分离研究的音频数据集，包含来自Smule应用的伴奏、孤立人声和混合音频，覆盖155个国家、36种语言、6456位艺术家和11494首作品，是真实世界的移动表演数据。