Convolutional Neural Networks and X-Vector Embedding for DCASE2018 Acoustic Scene Classification Challenge

article in a collection out of WoS and Scopus

English

In this paper, the Brno University of Technology (BUT) team submissions for Task 1 (Acoustic Scene Classification, ASC) of the DCASE-2018 challenge are described. Also, the analysis of different methods on the leaderboard set is provided. The proposed approach is a fusion of two different Convolutional Neural Network (CNN) topologies. The first one is the common two-dimensional CNNs which is mainly used in image classification. The second one is a one-dimensional CNN for extracting fixed-length audio segment embeddings, so called x-vectors, which has also been used in speech processing, especially for speaker recognition. In addition to the different topologies, two types of features were tested: log mel-spectrogram and CQT features. Finally, the outputs of different systems are fused using a simple output averaging in the best performing system. Our submissions ranked third among 24 teams in the ASC sub-task A (task 1a).

Audio scene classification, Convolutional neural networks, Deep learning, x-vectors, Regularized LDA

ZEINALI, H.; BURGET, L.; ČERNOCKÝ, J.

19. 11. 2018

Tampere University of Technology

Surrey

978-952-15-4262-6

Proceedings of DCASE 2018 Workshop

1

5

5

http://dcase.community/documents/workshop2018/proceedings/DCASE2018Workshop_Zeinali_149.pdf

zeinali_dcase2018_149.pdf