FSD50K: An Open Dataset of Human-Labeled Sound Events

IEEE/ACM Transactions on Audio Speech and Language Processing, Journal Year: 2021, Volume and Issue: 30, P. 829 - 852

Published: Dec. 9, 2021

Most existing datasets for sound event recognition (SER) are relatively small and/or domain-specific, with the exception of AudioSet, based on over 2 M tracks from YouTube videos and encompassing 500 classes. However, AudioSet is not an open dataset as its official release consists pre-computed audio features. Downloading original can be problematic due to gradually disappearing usage rights issues. To provide alternative benchmark thus foster SER research, we introduce FSD50K , containing 51 k clips totalling 100 h manually labeled using 200 classes drawn Ontology. The licensed under Creative Commons licenses, making freely distributable (including waveforms). We a detailed description FSD50K creation process, tailored particularities Freesound data, including challenges encountered solutions adopted. include comprehensive characterization along discussion limitations key factors allow audio-informed usage. Finally, conduct classification experiments baseline systems well insight main consider when splitting data SER. Our goal develop widely adopted by community new research.

Language: Английский

---

Computational bioacoustics with deep learning: a review and roadmap

PeerJ, Journal Year: 2022, Volume and Issue: 10, P. e13152 - e13152

Published: March 21, 2022

Animal vocalisations and natural soundscapes are fascinating objects of study, contain valuable evidence about animal behaviours, populations ecosystems. They studied in bioacoustics ecoacoustics, with signal processing analysis an important component. Computational has accelerated recent decades due to the growth affordable digital sound recording devices, huge progress informatics such as big data, machine learning. Methods inherited from wider field deep learning, including speech image processing. However, tasks, demands data characteristics often different those addressed or music analysis. There remain unsolved problems, tasks for which is surely present many acoustic signals, but not yet realised. In this paper I perform a review state art learning computational bioacoustics, aiming clarify key concepts identify analyse knowledge gaps. Based on this, offer subjective principled roadmap learning: topics that community should aim address, order make most future developments AI informatics, use audio answering zoological ecological questions.

Language: Английский

---

Drivers of fatal bird collisions in an urban center

Proceedings of the National Academy of Sciences, Journal Year: 2021, Volume and Issue: 118(24)

Published: June 7, 2021

Significance Collisions with built structures are an important source of bird mortality, killing hundreds millions birds annually in North America alone. Nocturnally migrating attracted to and disoriented by artificial lighting, making light pollution factor collision there is growing interest mitigating the impacts protect birds. We use two decades data show that migration magnitude, output, wind conditions predictors collisions at a large building Chicago decreasing lighted window area could reduce mortality ∼60%. Our finding extinguishing lights can death has global implications for conservation action campaigns aimed eliminating cause mortality.

Language: Английский

---

scikit‐maad: An open‐source and modular toolbox for quantitative soundscape analysis in Python

Methods in Ecology and Evolution, Journal Year: 2021, Volume and Issue: 12(12), P. 2334 - 2340

Published: Aug. 27, 2021

Abstract Passive acoustic monitoring is increasingly being applied to terrestrial, marine and freshwater environments, providing cost‐efficient methods for surveying biodiversity. However, processing the avalanche of audio recordings remains challenging, represents nowadays a major bottleneck that slows down its application in research conservation. We present scikit‐maad, an open‐source Python package dedicated analysis environmental recordings. This was designed (a) load process digital audio, (b) segment find regions interest, (c) compute features (d) estimate sound pressure levels. The also provides field comprehensive online documentation includes practical examples with step‐by‐step instructions beginners advanced users. scikit‐maad opens possibility efficiently scan large datasets easily integrate additional machine learning packages into analysis, allowing measure properties identify key patterns all kinds soundscapes. To support reproducible research, released under BSD licence, which allows unrestricted redistribution commercial private use. development will create synergies between community ecoacousticians, such as engineers, data scientists, ecologists, biologists conservation practitioners, explore understand processes underlying diversity ecological systems.

Language: Английский

---

Advances in automatic identification of flying insects using optical sensors and machine learning

Scientific Reports, Journal Year: 2021, Volume and Issue: 11(1)

Published: Jan. 15, 2021

Abstract Worldwide, farmers use insecticides to prevent crop damage caused by insect pests, while they also rely on pollinators enhance yield and other as natural enemies of pests. In order target pesticides pests only, must know exactly where when beneficial insects are present in the field. A promising solution this problem could be optical sensors combined with machine learning. We obtained around 10,000 records flying found oilseed rape ( Brassica napus ) crops, using an remote sensor evaluated three different classification methods for signals, reaching over 80% accuracy. demonstrate that it is possible classify flight, making optimize application space time. This will enable a technological leap precision agriculture, focus prudent environmentally-sensitive top priority.

Language: Английский

---

A review of automatic recognition technology for bird vocalizations in the deep learning era

Ecological Informatics, Journal Year: 2022, Volume and Issue: 73, P. 101927 - 101927

Published: Nov. 25, 2022

Language: Английский

---

Localization in wireless sensor networks and wireless multimedia sensor networks using clustering techniques

Multimedia Tools and Applications, Journal Year: 2023, Volume and Issue: 83(3), P. 6829 - 6879

Published: June 18, 2023

Language: Английский

---

Computational methods for detecting insect vibrational signals in field vibroscape recordings

Ecological Informatics, Journal Year: 2025, Volume and Issue: 86, P. 103003 - 103003

Published: Jan. 18, 2025

Language: Английский

---

Improving Smart Cities Safety Using Sound Events Detection Based on Deep Neural Network Algorithms

Informatics, Journal Year: 2020, Volume and Issue: 7(3), P. 23 - 23

Published: July 20, 2020

In recent years, security in urban areas has gradually assumed a central position, focusing increasing attention on citizens, institutions and political forces. Security problems have different nature—to name few, we can think of the deriving from citizens’ mobility, then move to microcrime, end up with ever-present risk terrorism. Equipping smart city an infrastructure sensors capable alerting managers about possible becomes crucial for safety citizens. The use unmanned aerial vehicles (UAVs) manage needs is now widespread, highlight risks public safety. These were increased using these devices carry out terrorist attacks various places around world. Detecting presence drones not simple procedure given small size only rotating parts. This study presents results studies carried detection UAVs outdoor/indoor sound environments. For UAVs, measuring emitted by algorithms based deep neural networks identifying their spectral signature that used. obtained suggest adoption this methodology improving cities.

Language: Английский

---

Ambient acoustic event assistive framework for identification, detection, and recognition of unknown acoustic events of a residence

Advanced Engineering Informatics, Journal Year: 2021, Volume and Issue: 47, P. 101238 - 101238

Published: Jan. 1, 2021

Language: Английский

---