Joint Analysis of Acoustic Events and Scenes Based on Multitask Learning
This work addresses the challenge of environmental sound analysis for applications like audio surveillance or smart environments, but it is incremental as it builds on existing neural network methods by integrating tasks.
The paper tackled the problem of separately analyzing acoustic events and scenes by proposing a multitask learning approach that jointly analyzes them, resulting in a 10.66 percentage point improvement in F-score for acoustic event detection compared to a conventional method.
Acoustic event detection and scene classification are major research tasks in environmental sound analysis, and many methods based on neural networks have been proposed. Conventional methods have addressed these tasks separately; however, acoustic events and scenes are closely related to each other. For example, in the acoustic scene `office', the acoustic events `mouse clicking' and `keyboard typing' are likely to occur. In this paper, we propose multitask learning for joint analysis of acoustic events and scenes, which shares the parts of the networks holding information on acoustic events and scenes in common. By integrating the two networks, we expect that information on acoustic scenes will improve the performance of acoustic event detection. Experimental results obtained using TUT Sound Events 2016/2017 and TUT Acoustic Scenes 2016 datasets indicate that the proposed method improves the performance of acoustic event detection by 10.66 percentage points in terms of the F-score, compared with a conventional method based on a convolutional recurrent neural network.