A Deep Learning Framework for Urban Ecoacoustic Monitoring and Management: A Synthetic-to-Real CNN-Transformer Approach

• Luis J. R. Oliveira UERN
• Ingrid G. C. Lira UFERSA
• Maximiliano A. da S. Lopes UERN

DOI: https://doi.org/10.5753/eniac.2026.30296

Resumo

Urban acoustic monitoring generates massive volumes of continuous data, making manual soundscape inspection unfeasible at scale. This work proposes a hybrid CNN-Transformer framework for urban ecoacoustic monitoring, modeling temporal interactions among Biophony, Geophony, and Anthropophony. A Transformer encoder-only was trained on probabilities extracted by PANNs from a controlled synthetic dataset (ESC-50), eliminating dependence on weak labels. The model achieved a macro F1-Score of 0.9960 and generalized to three real-world datasets, with a Spearman correlation of 0.54 (95% CI: [0.40, 0.65], p < 0.001, n = 127) between Anthropophony predictions and measured sound levels; predictions were spatialized via Ordinary Kriging, yielding a kriging RMSE of 0.07 in Madrid. Results show that the framework captures the dynamics of biophony suppression by anthropogenic noise and provides quantitative support for urban environmental management and biodiversity conservation.

Referências

Challéat, S., et al. (2020). Silent Cities: An acoustic monitoring network of the Covid-19 lockdown. CNRS. [link]

Correia, G., & Ventura, T. (2025). Assessing the Potential of Deep Learning in the Acoustic Identification of Threatened Birds in the Pantanal. In Anais do XXII Encontro Nacional de Inteligência Artificial e Computacional, (pp. 1491-1502). Porto Alegre: SBC. DOI: 10.5753/eniac.2025.12424

De Coensel, B., Sun, K., & Botteldooren, D. (2017). Urban Soundscapes of the World: Selection and reproduction of urban acoustic environments with soundscape in mind. In INTER-NOISE and NOISE-CON Congress and Conference Proceedings (Vol. 255, No. 2, pp. 5407-5413). Institute of Noise Control Engineering.

Isaaks, E. H., & Srivastava, R. M. (1989). An introduction to applied geostatistics. Oxford University Press.

Kong, Q., et al. (2020). PANNs: Large-scale pretrained audio neural networks for audio pattern recognition. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 28, 2880–2894.

Krause, B. L. (1993). The niche hypothesis. The Soundscape Newsletter, 6, 6–10.

McInnes, L., Healy, J., & Melville, J. (2018). UMAP: Uniform manifold approximation and projection for dimension reduction. Journal of Open Source Software, 3(29), 861. DOI: 10.21105/joss.00861

Mesaros, A., Heittola, T., Virtanen, T., & Plumbley, M. D. (2021). Sound event detection: A tutorial. IEEE Signal Processing Magazine, 38(5), 67–83. DOI: 10.1109/MSP.2021.3090678

Nery, S., Batista, F., & Bezerra, S. (2025). ColabSomGeo: Ferramenta de monitoramento de poluição sonora subsidiada por modelo de aprendizagem. In Anais Estendidos do XXXI Simpósio Brasileiro de Sistemas Multimídia e Web, (pp. 99-102). Porto Alegre: SBC. DOI: 10.5753/webmedia_estendido.2025.16410

Nikolenko, S. I. (2021). Synthetic data for deep learning (Vol. 174). Springer International Publishing.

Ooi, K., et al. (2021). SINGA:PURA: A polyphonic urban sound dataset with spatiotemporal context. In 2021 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference (APSIPA ASC) (pp. 463–470). IEEE.

Peralta, A., Santos, E., Xie, J., & Colonna, J. (2024). Recuperação semântica de paisagens sonoras usando banco de dados vetoriais. In Anais do XV Workshop de Computação Aplicada à Gestão do Meio Ambiente e Recursos Naturais, (pp. 51-60). Porto Alegre: SBC. DOI: 10.5753/wcama.2024.2316

Piczak, K. J. (2015). Environmental sound classification with convolutional neural networks. In 2015 IEEE International Workshop on Machine Learning for Signal Processing (MLSP) (pp. 1–6). IEEE.

Pijanowski, B. C., Villanueva-Rivera, L. J., Slabbekoorn, H., Farina, A., Krause, B. L., Napoletano, B. M., Gage, S. H., & Pieretti, N. (2011). Soundscape ecology: The science of sound in the landscape. BioScience, 61(3), 203–216. DOI: 10.1525/bio.2011.61.3.6

Quinn, C. A., et al. (2022). Soundscape classification with convolutional neural networks reveals temporal and geographic patterns in ecoacoustic data. Ecological Indicators, 138, 108831.

Ronchini, F., & Serizel, R. (2022). A benchmark of state-of-the-art sound event detection systems evaluated on synthetic soundscapes. In ICASSP 2022 – IEEE International Conference on Acoustics, Speech and Signal Processing (pp. 1031–1035). IEEE.

Salamon, J., & Bello, J. P. (2017). Deep convolutional neural networks and data augmentation for environmental sound classification. IEEE Signal Processing Letters, 24(3), 279–283.

Stowell, D. (2022). Computational bioacoustics with deep learning: A review and roadmap. PeerJ, 10, e13152.

Sugai, L. S. M., Silva, T. S. F., Ribeiro Jr., J. W., & Llusia, D. (2019). Terrestrial passive acoustic monitoring: Review and perspectives. BioScience, 69(1), 15–25. DOI: 10.1093/biosci/biy147

Vaswani, A., et al. (2017). Attention is all you need. In Advances in Neural Information Processing Systems (Vol. 30).