A large-scale Eurasian tree sparrow acoustic dataset with hierarchical song syllable annotations

Eurasian tree sparrow (Passer montanus) acoustic data from six sampling sites in Gansu, China, collected during the 2023 and 2024 breeding seasons. The dataset contains 64,811 original MP3 recordings with companion BirdNET detection tables, a 581-record confirmed-song index, and 138 prepared analysis excerpts from 116 original recordings. The excerpts contain 2,556 labelled syllables organized into 27 families and 41 terminal categories. Version 1.0.0 includes seven linked metadata tables, 138 Raven selection tables, a data dictionary, bilingual annotation and reuse guides, technical-validation materials and integrity manifests. The label-assessment inputs match the extraction intervals of all 2,556 released syllables. Known clock errors are flagged for 2,063 recordings. Syllable coordinates refer to the analysis WAVs; precise crop coordinates in original MP3s are not supplied. The distribution consists of 49 independent raw-audio archives, one analysis-audio archive and one metadata/annotation/validation archive, plus download instructions and checksums. Companion code version 1.0.0 supports inspection, sample-exact reconstruction, Raven export, matched assessments and figure reproduction. Dataset materials use CC BY 4.0; software and third-party materials retain their applicable licences. Companion code: https://doi.org/10.5281/zenodo.22879603. Code repository: https://github.com/linlin0329/tree-sparrow-acoustic-code.