Using Deep Learning to Classify Environmental Sounds in the Habitat of Western Black-Crested Gibbons DOI Creative Commons

Ruiqi Hu,

Kunrong Hu,

Leiguang Wang

и другие.

Diversity, Год журнала: 2024, Номер 16(8), С. 509 - 509

Опубликована: Авг. 22, 2024

The western black-crested gibbon (Nomascus concolor) is a rare and endangered primate that inhabits southern China northern Vietnam, has become key conservation target due to its distinctive call highly status, making identification monitoring particularly urgent. Identifying calls of the using passive acoustic data crucial method for studying analyzing these gibbons; however, traditional recognition models often overlook temporal information in audio features fail adapt channel-feature weights. To address issues, we propose an innovative deep learning model, VBSNet, designed recognize classify variety biological calls, including those gibbons certain bird species. model incorporates image feature extraction capability VGG16 convolutional network, sequence modeling bi-directional LSTM, selection SE attention module, realizing multimodal fusion image, information. In constructed dataset, VBSNet achieved best performance evaluation metrics accuracy, precision, recall, F1-score, accuracy 98.35%, demonstrating high generalization ability. This study provides effective field automated bioacoustic monitoring, which great theoretical practical significance supporting wildlife maintaining biodiversity.

Язык: Английский

Toward Inclusive Smart Cities: Sound-Based Vehicle Diagnostics, Emergency Signal Recognition, and Beyond DOI Creative Commons
Amr E. Eldin Rashed, Yousry AbdulAzeem, Tamer Ahmed Farrag

и другие.

Machines, Год журнала: 2025, Номер 13(4), С. 258 - 258

Опубликована: Март 21, 2025

Sound-based early fault detection for vehicles is a critical yet underexplored area, particularly within Intelligent Transportation Systems (ITSs) smart cities. Despite the clear necessity sound-based diagnostic systems, scarcity of specialized publicly available datasets presents major challenge. This study addresses this gap by contributing in multiple dimensions. Firstly, it emphasizes significance diagnostics real-time faults through analyzing sounds directly generated vehicles, such as engine or brake noises, and classification external emergency sounds, like sirens, relevant to vehicle safety. Secondly, paper introduces novel dataset encompassing environmental noises specifically curated address absence datasets. A comprehensive framework proposed, combining audio preprocessing, feature extraction (via Mel Spectrograms, MFCCs, Chromatograms), using 11 models. Evaluations both compact (52 features) expanded (126 representations show that several classes (e.g., Engine Misfire, Fuel Pump Cartridge Fault, Radiator Fan Failure) achieve near-perfect accuracy, though acoustically similar Universal Joint Failure, Knocking, Pre-ignition Problem remain challenging. Logistic Regression yielded highest accuracy 86.5% (DB1) features, while neural networks performed best DB2 DB3, achieving 88.4% 85.5%, respectively. In second scenario, Bayesian-Optimized Weighted Soft Voting with Feature Selection (BOWSVFS) approach significantly enhancing 91.04% DB1, 88.85% DB2, 86.85% DB3. These results highlight effectiveness proposed methods addressing key ITS limitations accessibility individuals disabilities auditory-based recognition systems.

Язык: Английский

Процитировано

0
In the songs of Hainan gibbons: Automated individual dynamic monitoring from acoustic recordings DOI
Zidi Wang,

Haigang Ma,

Xukai Zhong

и другие.

Biological Conservation, Год журнала: 2024, Номер 294, С. 110634 - 110634

Опубликована: Май 22, 2024

Язык: Английский

Процитировано

1
Meta-Embedded Clustering (MEC): A new method for improving clustering quality in unlabeled bird sound datasets DOI Creative Commons
Joachim POUTARAUD, Jérôme Sueur,

Christophe Thébaud

и другие.

Ecological Informatics, Год журнала: 2024, Номер 82, С. 102687 - 102687

Опубликована: Июнь 15, 2024

In recent years, ecoacoustics has offered an alternative to traditional biodiversity monitoring techniques with the development of passive acoustic (PAM) systems allowing, among others, detect and identify species that are difficult by human observers, automatically. PAM typically generate large audio datasets, but using these infer ecologically meaningful information remains challenging. most cases, several thousand hours recordings need be manually labeled experts limiting operability systems. Based on developments meta-learning algorithms unsupervised learning techniques, we propose here Meta-Embedded Clustering (MEC), a new method high potential for improving clustering quality in unlabeled bird sound datasets. MEC is organized two main steps, with: (a) fine-tuning pretrained convolutional neural network (CNN) backbone different pseudo-labeled data, (b) manually-labeled sounds latent space based vector embeddings extracted from fine-tuned CNN. The significantly enhanced average performance less than 1% more 80%, greatly outperforming approach relying solely CNN features general neotropical database. However, this came cost excluding portion data categorized as noise. By should facilitate work ecoacousticians managing units song/call clustered according their similarities, identifying clusters undetected approaches.

Язык: Английский

Процитировано

0
Applying machine learning to primate bioacoustics: Review and perspectives DOI Creative Commons
Jules Cauzinille, Benoît Favre, Ricard Marxer

и другие.

American Journal of Primatology, Год журнала: 2024, Номер 86(10)

Опубликована: Авг. 9, 2024

Abstract This paper provides a comprehensive review of the use computational bioacoustics as well signal and speech processing techniques in analysis primate vocal communication. We explore potential implications machine learning deep methods, from simple supervised algorithms to more recent self‐supervised models, for analyzing large data sets obtained within emergence passive acoustic monitoring approaches. In addition, we discuss importance automated vocalization tackling essential questions on animal communication highlighting role comparative linguistics bioacoustic research. also examine challenges associated with collection annotation provide insights into solutions. Overall, this runs through set common or innovative perspectives applications outlines opportunities future research rapidly developing field.

Язык: Английский

Процитировано

0
Benchmarking for the automated detection and classification of southern yellow-cheeked crested gibbon calls from passive acoustic monitoring data DOI Creative Commons
Dena J. Clink,

Hope Cross-Jaya,

Jinsung Kim

и другие.

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2024, Номер unknown

Опубликована: Авг. 19, 2024

Recent advances in deep and transfer learning have revolutionized our ability for the automated detection classification of acoustic signals from long-term recordings. Here, we provide a benchmark southern yellow-cheeked crested gibbon ( Nomascus gabriellae ) calls collected using autonomous recording units (ARUs) Andoung Kraleung Village, Cambodia. We compared performance support vector machines (SVMs), quasi-DenseNet architecture (Koogu), with pretrained convolutional neural network (ResNet50) models trained on ‘ImageNet’ dataset, embeddings global birdsong model (BirdNET) based an EfficientNet architecture. also investigated impact varying number training samples these models. found that BirdNET had superior smaller samples, whereas Koogu ResNet50 only acceptable larger (>200 samples). Effective approaches are critical monitoring endangered species, like gibbons. It is unclear how generalizable results other signals, future work vocal species will be informative. Code data publicly available benchmarking.

Язык: Английский

Процитировано

0
Using Deep Learning to Classify Environmental Sounds in the Habitat of Western Black-Crested Gibbons DOI Creative Commons

Ruiqi Hu,

Kunrong Hu,

Leiguang Wang

и другие.

Diversity, Год журнала: 2024, Номер 16(8), С. 509 - 509

Опубликована: Авг. 22, 2024

The western black-crested gibbon (Nomascus concolor) is a rare and endangered primate that inhabits southern China northern Vietnam, has become key conservation target due to its distinctive call highly status, making identification monitoring particularly urgent. Identifying calls of the using passive acoustic data crucial method for studying analyzing these gibbons; however, traditional recognition models often overlook temporal information in audio features fail adapt channel-feature weights. To address issues, we propose an innovative deep learning model, VBSNet, designed recognize classify variety biological calls, including those gibbons certain bird species. model incorporates image feature extraction capability VGG16 convolutional network, sequence modeling bi-directional LSTM, selection SE attention module, realizing multimodal fusion image, information. In constructed dataset, VBSNet achieved best performance evaluation metrics accuracy, precision, recall, F1-score, accuracy 98.35%, demonstrating high generalization ability. This study provides effective field automated bioacoustic monitoring, which great theoretical practical significance supporting wildlife maintaining biodiversity.

Язык: Английский

Процитировано

0