SONABIO PAM Dataset — Processed Results

A passive acoustic monitoring survey conducted in the Colombian Amazon collected 722 hours of tropical soundscape data across 38 recorder deployments over 37 days, generating over 43,000 audio files. The survey was carried out with a participatory approach involving fourteen families and a local school, in support of community forestry initiatives. Processing through pamflow covered five pipelines: Data preparation: structured the media and deployment metadata according to pamDP data standards (documentation) Quality control: produced spatial, temporal, and audio diagnostics Species detection: automatically identified 349 bird species from over 347,000 detections without a predefined target-species list. Acoustic indices: computed 10 acoustic indices, including ACI, ADI, BI, NP, Hf, Ht, H, NDSI, NP, RMS, SC. Graphical soundscape: computed a spectro-temporal representation of the acoustics of each deployment location that can be used to understand global dynamics of the soundscape. The field audio recordings and monitoring design were co-created with the Humboldt Institute and the local community. If you use this dataset for testing or for developing other open-source tools, please be sure to acknowledge all participating community members, researchers, and institutions involved. The data cannot be used for machine learning purposes or for training AI models without explicit consent from the data creators. Moises Romero Franco, Anibal De Jesus Martinez Peña, Daniel Chitiva Diaz, Nubia Quintero Medina, Edgar Gustavo Rodriguez Garzon, Granja Escuela Capricho, Maria De Los Angeles Gaitan Muños, Flor Marleni Gaitan Muños, Jefferson Cocinero Garcia, Jose Eduardo Leon Triana, Cesar Augusto Urrego Gonzalez, Anderson Urrea Tacuma, Luz Idalba Restrepo Suarez, Maria De Los Angeles Gaitan Muñoz, Edwin Alcides Reyes Lopez, Juan Sebastián Ulloa, Juan Sebastián Cañas, Marcela González, Raúl Gómez.