2020/11/02 by Nicolas Turpault, Turpault, Nicolas, Scott Wisdom +12 · 1 citation
Computer Science · #Audio and Speech Processing (eess.AS) #FOS: Computer and information sciences #FOS: Electrical engineering #Music Technology and Sound Studies #Music and Audio Processing #Sound (cs.SD) #Speech and Audio Processing #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.2011.00801
openalex publication_date 2020/11/02 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
We propose a benchmark of state-of-the-art sound event detection systems\n(SED). We designed synthetic evaluation sets to focus on specific sound event\ndetection challenges. We analyze the performance of the submissions to DCASE\n2021 task 4 depending on time related modifications (time position of an event\nand length of clips) and we study the impact of non-target sound events and\nreverberation. We show that the localization in time of sound events is still a\nproblem for SED systems. We also show that reverberation and non-target sound\nevents are severely degrading the performance of the SED systems. In the latter\ncase, sound separation seems like a promising solution.\n