Stacked Convolutional and Recurrent Neural Networks for Bird Audio Detection
Sound
2017-06-08 v1 Machine Learning
Abstract
This paper studies the detection of bird calls in audio segments using stacked convolutional and recurrent neural networks. Data augmentation by blocks mixing and domain adaptation using a novel method of test mixing are proposed and evaluated in regard to making the method robust to unseen data. The contributions of two kinds of acoustic features (dominant frequency and log mel-band energy) and their combinations are studied in the context of bird audio detection. Our best achieved AUC measure on five cross-validations of the development data is 95.5% and 88.1% on the unseen evaluation data.
Cite
@article{arxiv.1706.02047,
title = {Stacked Convolutional and Recurrent Neural Networks for Bird Audio Detection},
author = {Sharath Adavanne and Konstantinos Drossos and Emre Çakır and Tuomas Virtanen},
journal= {arXiv preprint arXiv:1706.02047},
year = {2017}
}
Comments
Accepted for European Signal Processing Conference 2017