Automatic acoustic detection of birds through deep learning: the first Bird Audio Detection challenge

Abstract

Assessing the presence and abundance of birds is important for monitoring specific species as well as overall ecosystem health. Many birds are most readily detected by their sounds, and thus passive acoustic monitoring is highly appropriate. Yet acoustic monitoring is often held back by practical limitations such as the need for manual configuration, reliance on example sound libraries, low accuracy, low robustness, and limited ability to generalise to novel acoustic conditions. Here we report outcomes from a collaborative data challenge showing that with modern machine learning including deep learning, general-purpose acoustic bird detection can achieve very high retrieval rates in remote monitoring data --- with no manual recalibration, and no pre-training of the detector for the target species or the acoustic conditions in the target environment. Multiple methods were able to attain performance of around 88% AUC (area under the ROC curve), much higher performance than previous general-purpose methods. We present new acoustic monitoring datasets, summarise the machine learning techniques proposed by challenge teams, conduct detailed performance evaluation, and discuss how such approaches to detection can be integrated into remote monitoring projects.

References in corpus (3)

• Convolutional Recurrent Neural Networks for Polyphonic Sound Event Detection
• Automatic large-scale classification of bird sounds is strongly improved by unsupervised feature learning
• Densely Connected CNNs for Bird Audio Detection

Cited by in corpus (27)

• Seeing biodiversity: perspectives in machine learning for wildlife conservation
• Machine Learning and Deep Learning -- A review for Ecologists
• Computational bioacoustics with deep learning: a review and roadmap
• Self-Supervised Speech Representation Learning: A Review
• WILDS: A Benchmark of in-the-Wild Distribution Shifts
• Global birdsong embeddings enable superior transfer learning for bioacoustic classification
• Robust sound event detection in bioacoustic sensor networks
• Weakly Labelled AudioSet Tagging with Attention Neural Networks
• Learning to detect an animal sound from five examples
• Multiscale CNN based Deep Metric Learning for Bioacoustic Classification: Overcoming Training Data Scarcity Using Dynamic Triplet Loss
• DCASE 2018 Challenge Surrey Cross-Task convolutional neural network baseline
• LEAF: A Learnable Frontend for Audio Classification
• Multimodal Self-Supervised Learning of General Audio Representations
• DECAR: Deep Clustering for learning general-purpose Audio Representations
• pykanto: a python library to accelerate research on wild bird song
• Learning audio representations via phase prediction
• A Bird Song Detector for improving bird identification through Deep Learning: a case study from Doñana
• Edge Intelligence for Wildlife Conservation: Real-Time Hornbill Call Classification Using TinyML
• Low Resource Species Agnostic Bird Activity Detection
• An Investigation of the Effectiveness of Phase for Audio Classification
• Detection of Audio-Video Synchronization Errors Via Event Detection
• Recognizing bird species in diverse soundscapes under weak supervision
• CNN-based Discriminative Training for Domain Compensation in Acoustic Event Detection with Frame-wise Classifier
• Towards Learning Universal Audio Representations
• Modelling Animal Biodiversity Using Acoustic Monitoring and Deep Learning
• Classification of animal sounds in a hyperdiverse rainforest using Convolutional Neural Networks
• Training Set Synthesis for Bioacoustic Denoising: A Case Study With Mice