Name
Abstract | ||
|---|---|---|
Singing voice separation aims to separate music into vocals and accompaniment components. One of the major constraints for the task is the limited amount of training data with separated vocals. Data augmentation techniques such as random source mixing have been shown to make better use of existing data and mildly improve model performance. We propose a novel data augmentation technique, chromagram-based pitch-aware remixing, where music segments with high pitch alignment are mixed. By performing controlled experiments in both supervised and semi-supervised settings, we demonstrate that training models with pitch-aware remixing significantly improves the test signal-to-distortion ratio (SDR). |
Year | DOI | Venue |
|---|---|---|
2022 | 10.1109/ICASSP43922.2022.9747612 | ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) |
Keywords | DocType | ISSN |
Singing voice separation,augmentation,pitch-aware,chromagram,self-training | Conference | 1520-6149 |
ISBN | Citations | PageRank |
978-1-6654-0541-6 | 0 | 0.34 |
References | Authors | |
6 | 7 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Siyuan Yuan | 1 | 0 | 0.34 |
| Zhepei Wang | 2 | 3 | 0.78 |
| Umut Isik | 3 | 10 | 3.33 |
| Ritwik Giri | 4 | 164 | 12.93 |
| J.-M. Valin | 5 | 740 | 66.29 |
| Michael M. Goodwin | 6 | 0 | 0.34 |
| Arvindh Krishnaswamy | 7 | 12 | 3.37 |