1 paper · 1 filter
Azalea Gui, Woosung Choi, Junghyun Koo +5
The performance of deep learning models for music source separation heavily depends on training data quality. However, datasets are often corrupted by difficult-to-detect artifacts…