Historical report; not the current benchmark. This document retains original claims and scores, some of which were corrected or withdrawn. Use the current study and research record for the final interpretation.
New Baseline Results (tiered)
macro-F1 reported as single-split (seed 42) and mean ± std over 10 stratified splits. WhAM/AVES report best layer per task.
Feature floors
| Model | Unit | Coda Type | Individual ID |
|---|---|---|---|
| Raw ICI | 0.599 / 0.584±0.013 | 0.931 / 0.908±0.057 | 0.493 / 0.467±0.022 |
| Raw Mel | 0.740 / 0.765±0.029 | 0.097 / 0.135±0.023 | 0.280 / 0.287±0.041 |
Strong baselines (label-free)
| Model | Unit | Coda Type | Individual ID |
|---|---|---|---|
| AVES | 0.905 / 0.916±0.011 (L8) | 0.290 / 0.237±0.042 (L4) | 0.418 / 0.402±0.054 (L7) |
| BirdNET | 0.863 / 0.897±0.017 | 0.272 / 0.231±0.035 | 0.322 / 0.402±0.042 |
| Perch | pending | pending | pending |
SOTA integrated baseline
| Model | Unit | Coda Type | Individual ID |
|---|---|---|---|
| WhAM | 0.895 / 0.909±0.015 (L19) | 0.261 / 0.227±0.033 (L19) | 0.426 / 0.394±0.083 (L7) |
Cell format: single-split / mean±std (best layer).