Towards duration robust weakly supervised sound event detection

TL;DR

This paper introduces a duration robust CRNN framework for weakly-supervised sound event detection that performs well without prior duration knowledge, especially on datasets with short events, using novel post-processing and data augmentation techniques.

Contribution

The paper proposes a new CRNN-based model with a Triple Threshold post-processing strategy and data augmentation methods, improving localization performance in weakly-supervised SED without needing duration labels.

Findings

Outperforms existing methods on DCASE2018 and URBAN-SED datasets.

Achieves similar performance to supervised models on URBAN-SED.

Post-processing significantly reduces localization performance drop.

Abstract

Sound event detection (SED) is the task of tagging the absence or presence of audio events and their corresponding interval within a given audio clip. While SED can be done using supervised machine learning, where training data is fully labeled with access to per event timestamps and duration, our work focuses on weakly-supervised sound event detection (WSSED), where prior knowledge about an event's duration is unavailable. Recent research within the field focuses on improving segment- and event-level localization performance for specific datasets regarding specific evaluation metrics. Specifically, well-performing event-level localization requires fully labeled development subsets to obtain event duration estimates, which significantly benefits localization performance. Moreover, well-performing segment-level localization models output predictions at a coarse-scale (e.g., 1 second), hindering their deployment on datasets containing very short events (< 1 second). This work proposes a duration robust CRNN (CDur) framework, which aims to achieve competitive performance in terms of segment- and event-level localization. This paper proposes a new post-processing strategy named "Triple Threshold" and investigates two data augmentation methods along with a label smoothing method within the scope of WSSED. Evaluation of our model is done on the DCASE2017 and 2018 Task 4 datasets, and URBAN-SED. Our model outperforms other approaches on the DCASE2018 and URBAN-SED datasets without requiring prior duration knowledge. In particular, our model is capable of similar performance to strongly-labeled supervised models on the URBAN-SED dataset. Lastly, ablation experiments to reveal that without post-processing, our model's localization performance drop is significantly lower compared with other approaches.