DCASE 2018 Challenge Surrey Cross-Task convolutional neural network baseline

TL;DR

This paper presents a unified CNN baseline for all five DCASE 2018 audio classification and sound event detection tasks, demonstrating the effectiveness of deeper networks across most tasks.

Contribution

It introduces a cross-task CNN baseline system based on 4- and 8-layer networks from AlexNet and VGG, providing a common framework for diverse audio tasks.

Findings

Deeper CNN (8 layers) outperforms shallower CNN (4 layers) on most tasks.

Achieved competitive accuracy and F1 scores across all five tasks.

Released open-source code to facilitate future research.

Abstract

The Detection and Classification of Acoustic Scenes and Events (DCASE) consists of five audio classification and sound event detection tasks: 1) Acoustic scene classification, 2) General-purpose audio tagging of Freesound, 3) Bird audio detection, 4) Weakly-labeled semi-supervised sound event detection and 5) Multi-channel audio classification. In this paper, we create a cross-task baseline system for all five tasks based on a convlutional neural network (CNN): a "CNN Baseline" system. We implemented CNNs with 4 layers and 8 layers originating from AlexNet and VGG from computer vision. We investigated how the performance varies from task to task with the same configuration of neural networks. Experiments show that deeper CNN with 8 layers performs better than CNN with 4 layers on all tasks except Task 1. Using CNN with 8 layers, we achieve an accuracy of 0.680 on Task 1, an accuracy of 0.895 and a mean average precision (MAP) of 0.928 on Task 2, an accuracy of 0.751 and an area under the curve (AUC) of 0.854 on Task 3, a sound event detection F1 score of 20.8% on Task 4, and an F1 score of 87.75% on Task 5. We released the Python source code of the baseline systems under the MIT license for further research.