Acoustic scene classification: from a hybrid classifier to deep learning

De Montfort University Open Research Archive

Show simple item record

dc.contributor.author Vafeiadis, Anastasios en
dc.contributor.author Kalatzis, Dimitrios en
dc.contributor.author Votis, Konstantinos en
dc.contributor.author Giakoumis, Dimitrios en
dc.contributor.author Tzovaras, Dimitrios en
dc.contributor.author Chen, Liming en
dc.contributor.author Hamzaoui, Raouf en
dc.date.accessioned 2017-12-12T09:36:50Z
dc.date.available 2017-12-12T09:36:50Z
dc.date.issued 2017-11-16
dc.identifier.citation Vafeiadis, A. et al. (2017) Acoustic scene classification: from a hybrid classifier to deep learning. In: Proceedings of the Detection and Classification of Acoustic Scenes and Events 2017 Workshop (DCASE2017), Munich, Nov. 2017. en
dc.identifier.isbn 9789521540424
dc.identifier.uri http://hdl.handle.net/2086/15000
dc.description.abstract This report describes our contribution to the 2017 Detection and Classification of Acoustic Scenes and Events (DCASE) challenge. We investigated two approaches for the acoustic scene classification task. Firstly, we used a combination of features in the time and frequency domain and a hybrid Support Vector Machines - Hidden Markov Model (SVM-HMM) classifier to achieve an average accuracy over 4-folds of 80.9% on the development dataset and 61.0% on the evaluation dataset. Secondly, by exploiting dataaugmentation techniques and using the whole segment (as opposed to splitting into sub-sequences) as an input, the accuracy of our CNN system was boosted to 95.9%. However, due to the small number of kernels used for the CNN and a failure of capturing the global information of the audio signals, it achieved an accuracy of 49.5% on the evaluation dataset. Our two approaches outperformed the DCASE baseline method, which uses log-mel band energies for feature extraction and a Multi-Layer Perceptron (MLP) to achieve an average accuracy over 4-folds of 74.8%. en
dc.language.iso en en
dc.subject Acoustic scene classification en
dc.subject feature extraction en
dc.subject deep learning en
dc.subject spectral features en
dc.subject data augmentation en
dc.title Acoustic scene classification: from a hybrid classifier to deep learning en
dc.type Conference en
dc.researchgroup CIIRG en
dc.peerreviewed Yes en
dc.funder EU Horizon 2020 en
dc.projectid Marie Skłodowska-Curie grant agreement No. 676157, project ACROSSING en
dc.cclicence CC-BY-NC en
dc.date.acceptance 2017-10-15 en


Files in this item

This item appears in the following Collection(s)

Show simple item record