Unsupervised domain adaptation for acoustic scene classification using band-wise statistics matching

Mezza AI, Habets E, Müller M, Sarti A (2021)

Publication Type: Conference contribution

Publication year: 2021

Publisher: European Signal Processing Conference, EUSIPCO

Book Volume: 2021-January

Pages Range: 11-15

Conference Proceedings Title: European Signal Processing Conference

Event location: Amsterdam NL

ISBN: 9789082797053

DOI: 10.23919/Eusipco47968.2020.9287533

Abstract

The performance of machine learning algorithms is known to be negatively affected by possible mismatches between training (source) and test (target) data distributions. In fact, this problem emerges whenever an acoustic scene classification system which has been trained on data recorded by a given device is applied to samples acquired under different acoustic conditions or captured by mismatched recording devices. To address this issue, we propose an unsupervised domain adaptation method that consists of aligning the first- and second-order sample statistics of each frequency band of target-domain acoustic scenes to the ones of the source-domain training dataset. This approach is devised to adapt audio samples from unseen devices before they are fed to a pre-trained classifier, thus avoiding any further learning phase. Using the DCASE 2018 Task 1-B development dataset, we show that the proposed method outperforms the state-of-the-art unsupervised methods found in the literature in terms of both source- and target-domain classification accuracy.

Authors with CRIS profile

Emanuël Habets Professur für wahrnehmungsbasierte räumliche Audiosignalverarbeitung (AudioLabs) (Stiftungsprofessur) Meinard Müller International Audio Laboratories Erlangen (AudioLabs)

Involved external institutions

Politecnico di Milano IT Italy (IT)

How to cite

APA:

Mezza, A.I., Habets, E., Müller, M., & Sarti, A. (2021). Unsupervised domain adaptation for acoustic scene classification using band-wise statistics matching. In European Signal Processing Conference (pp. 11-15). Amsterdam, NL: European Signal Processing Conference, EUSIPCO.

MLA:

Mezza, Alessandro Ilic, et al. "Unsupervised domain adaptation for acoustic scene classification using band-wise statistics matching." Proceedings of the 28th European Signal Processing Conference, EUSIPCO 2020, Amsterdam European Signal Processing Conference, EUSIPCO, 2021. 11-15.

BibTeX: Download