TreeVibes Provenance-Aware Validation Cohort (TV-PAVC v1): Manifests, Fixed Folds, and Reproducibility Artifacts

TV-PAVC v1 is a manifest-defined evaluation and reproducibility resource for the public TreeVibes vibroacoustic borer dataset. It defines a primary cohort of 2,072 deduplicated field clips from 24 nominal folders mapped to 23 recovered provenance/dependency-control groups, six fixed provenance-grouped cross-validation folds, analytical roles for non-primary material, deterministic reconstruction/verification code, frozen derived predictions/statistics, and release-verification evidence. No TreeVibes source audio is redistributed. A fresh verification run completed on 11 August 2026 matched all 4,154 field files by raw-file and decoded-PCM SHA-256, reproduced the 2,072-row primary manifest and 23 provenance groups, and detected no field-file or primary-manifest drift. Laboratory verification in that pass was limited to count/path parity. The 23 provenance groups are dependency-control units and are not asserted to be 23 independently verified biological trees.