2IoSR - Institute of Sound Recording (Department of Music and Sound Recording University of Surrey Guildford GU2 7XH Surrey United Kingdom - United Kingdom)
Abstract : The Signal Separation Evaluation Campaign (SiSEC) is a large-scale regular event aimed at evaluating current progress in source separation through a systematic and reproducible comparison of the participants’ algorithms, providing the source separation community with an invaluable glimpse of recent achievements and open challenges. This paper focuses on the music separation task from SiSEC 2018, which compares algorithms aimed at recovering instrument stems from a stereo mix. In this context, we conducted a subjective evaluation whereby 34 listeners picked which of six competing algorithms, with high objective performance scores, best separated the singing-voice stem from 13 professionally mixed songs. The subjective results reveal strong differences between the algorithms, and highlight the presence of song-dependent performance for state-of-the-art systems. Correlations between the subjective results and the scores of two popular performance metrics are also presented.
https://hal.inria.fr/hal-01945362
Contributor : Antoine Liutkus <>
Submitted on : Wednesday, December 5, 2018 - 11:45:14 AM Last modification on : Thursday, December 10, 2020 - 11:52:05 AM Long-term archiving on: : Wednesday, March 6, 2019 - 2:08:06 PM
Dominic Ward, Russel D. Mason, Chungeun Kim, Fabian-Robert Stöter, Antoine Liutkus, et al.. SiSEC 2018: State of the art in musical audio source separation - subjective selection of the best algorithm. WIMP: Workshop on Intelligent Music Production, Sep 2018, Huddersfield, United Kingdom. ⟨hal-01945362⟩