Semi-Supervised Singing Voice Separation with Noisy Self-Training
arXiv:2102.07961
Abstract
Recent progress in singing voice separation has primarily focused on supervised deep learning methods. However, the scarcity of ground-truth data with clean musical sources has been a problem for long. Given a limited set of labeled data, we present a method to leverage a large volume of unlabeled data to improve the model's performance. Following the noisy self-training framework, we first train a teacher network on the small labeled dataset and infer pseudo-labels from the large corpus of unlabeled mixtures. Then, a larger student network is trained on combined ground-truth and self-labeled datasets. Empirical results show that the proposed self-training scheme, along with data augmentation methods, effectively leverage the large unlabeled corpus and obtain superior performance compared to supervised methods.
arXiv commentAccepted at 2021 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2021)
References in corpus (8)
- Improved Noisy Student Training for Automatic Speech Recognition
- Music Source Separation in the Waveform Domain
- Voice Separation with an Unknown Number of Multiple Speakers
- Demucs: Deep Extractor for Music Sources with extra unlabeled data remixed
- Singing voice separation: a study on training data
- Voice and accompaniment separation in music using self-attention convolutional neural network
- Bootstrapping deep music separation from primitive auditory grouping principles
- Revisiting Representation Learning for Singing Voice Separation with Sinkhorn Distances