results/scores/antalia-2-mini.jsonl was still the 1.0.0 run under the 1.0.1 name, and results/timing/antalia-2-mini.json was missing. With these two files aggregate.py reproduces results/results.json. The scores come from the 10 October 2026 render of 1.0.1; the timing comes from a rerun of the same seeds after the output filter was cached, which is what the published 1.0.1 wheel contains. Audio identity between those two passes was not compared.
Alania-2 production row: best of 11 voices (12 Oct 2026, alania-v2.1)
Re-ran Alania-2 through the production API for all 11 built-in voices with the same harness and scoring; the row shows the best voice per column (WER/CER: Öncel, clean/UTMOS: Lina), named, with a footnote. Adds those two voices' per-clip scores.
Add Antalia-2 Mini 1.0.0 (default settings, 3 takes, one L4)
