Acoustic scene classification with fully convolutional neural networks and i-vectors

Autoren Matthias Dorfer
Bernhard Lehner
Hamid Eghbal-zadeh
Christoph Heindl
Fabian Paischer
Gerhard Widmer
Editoren
Titel Acoustic scene classification with fully convolutional neural networks and i-vectors
Typ sonst
Ort DCASE2018 Challange - IEEE AASP Challange on Detection and Classification of Acoustic Scenes and Events, March 30 - July 31, 2018
Monat July
Jahr 2018
Seiten http://dcase.community/challenge2018/task-acoustic-scene-classification-results-a#technical-reports
SCCH ID# 18098
Abstract

This technical report describes the CP-JKU team’s submissions for Task 1 - Subtask A (Acoustic Scene Classification, ASC) of the DCASE-2018 challenge. Our approach is still related to the methodology that achieved ranks 1 and 2 in the 2016 ASC challenge: a fusion of i-vector modelling using MFCC features derived from left and right audio channels, and deep convolutional neural networks (CNNs) trained on spectrograms. However, for our 2018 submission we have put a stronger focus on tuning and pushing the performance of our CNNs. The result of our experiments is a classification system that achieves classification accuracies of around 80% on the public Kaggle-Leaderboard.