End-to-End Sound Source Separation Conditioned On Instrument Labels

TitleEnd-to-End Sound Source Separation Conditioned On Instrument Labels
Publication TypeConference Paper
Year of Publication2019
Conference Name2019 International Conference on Acoustics, Speech, and Signal Processing
AuthorsSlizovskaia, O., Kim L., Haro G., & Gómez E.
Conference Start Date12/05/2019
Conference LocationBrighton, United Kingdom
AbstractCan we perform an end-to-end sound source separation with a variable number of sources using a deep learning model? This paper presents an extension of the Wave-U-Net model which allows end-to-end monaural source separation with a non-fixed number of sources. Furthermore, we propose multiplicative conditioning with instrument labels at the bottleneck of the Wave-U-Net and show its effect on the separation results. This approach can be further extended to other types of conditioning such as audio-visual and score-informed source separation.
preprint/postprint documenthttps://arxiv.org/abs/1811.01850