Weakly-supervised audio event detection using event-specific Gaussian filters and fully convolutional networks
Authors:
- 1 March 2017
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 791-795
Abstract
Audio event detection aims at discovering the elements inside an audio clip. In addition to labeling the clips with the audio events, we want to find out the temporal locations of these events. However, creating clearly annotated training data can be time-consuming. Therefore, we provide a model based on convolutional neural networks that relies only on weakly-supervised data for training. These data can be directly obtained from online platforms, such as Freesound, with the clip-level labels assigned by the uploaders. The structure of our model is extended to a fully convolutional networks, and an event-specific Gaussian filter layer is designed to advance its learning ability. Besides, this model is able to detect frame-level information, e.g., the temporal position of sounds, even when it is trained merely with clip-level labels.Keywords
This publication has 14 references indexed in Scilit:
- Event Localization in Music Auto-tagging2016
- Classification of breath and snore sounds using audio data recorded with smartphones in the home environment2016
- Cross-acoustic transfer learning for sound event classification2016
- Deep neural networks for automatic detection of screams and shouted speech in subway trains2016
- Is object localization for free? - Weakly-supervised learning with convolutional neural networks2015
- Fully convolutional networks for semantic segmentation2015
- Unsupervised feature learning for urban sound classification2015
- A Dataset and Taxonomy for Urban Sound Research2014
- End-to-end learning for music audio2014
- The use of the area under the ROC curve in the evaluation of machine learning algorithmsPattern Recognition, 1997