AudioSlots: A slot-centric generative model for audio separation
Abstract
In a range of recent works, object-centric architectures have been shown to be suitable for unsupervised scene decomposition in the vision domain. Inspired by these methods we present AudioSlots, a slot-centric generative model for blind source separation in the audio domain. AudioSlots is built using permutation-equivariant encoder and decoder networks. The encoder network based on the Transformer architecture learns to map a mixed audio spectrogram to an unordered set of independent source embeddings. The spatial broadcast decoder network learns to generate the source spectrograms from the source embeddings. We train the model in an end-to-end manner using a permutation invariant loss function. Our results on Libri2Mix speech separation constitute a proof of concept that this approach shows promise. We discuss the results and limitations of our approach in detail, and further outline potential ways to overcome the limitations and directions for future work.
- Publication:
-
arXiv e-prints
- Pub Date:
- May 2023
- DOI:
- 10.48550/arXiv.2305.05591
- arXiv:
- arXiv:2305.05591
- Bibcode:
- 2023arXiv230505591R
- Keywords:
-
- Computer Science - Sound;
- Computer Science - Computer Vision and Pattern Recognition;
- Electrical Engineering and Systems Science - Audio and Speech Processing
- E-Print:
- Accepted at the Self-supervision in Audio, Speech and Beyond (SASB) Workshop at ICASSP 2023