Take audit: 4_ZqJpLqwbVZwwVegf-YF

2026-08-22T09:37:31.412Z — Mika

0/13 judged · keyboard: Space play/pause, ↑/↓ focus, P pass, T Translation failure, F fail

Do a blind check (transcripts hidden)

Ground Truth

What the Tester actually said. Never pre-filled from an Engine — type what you hear.


Transcripts

mini_prod_baseline gpt-4o-mini-transcribe · degraded · en · bare — MER accuracy pending Ground Truth · 843 ms · verdict: unjudged

老師,I want a juice, too, please.

mini_auto_bare_deg gpt-4o-mini-transcribe · degraded · auto · bare — MER accuracy pending Ground Truth · 860 ms · verdict: unjudged

老師,I want a juice, too, please.

mini_auto_bare_clean gpt-4o-mini-transcribe · clean · auto · bare — MER accuracy pending Ground Truth · 642 ms · verdict: unjudged

老師,I want a just to face.

mini_auto_prompt_deg gpt-4o-mini-transcribe · degraded · auto · mixed prompt — MER accuracy pending Ground Truth · 686 ms · verdict: unjudged

老師,I want a juice, too, please.

mini_auto_prompt_clean gpt-4o-mini-transcribe · clean · auto · mixed prompt — MER accuracy pending Ground Truth · 708 ms · verdict: unjudged

老師,I want a just to face.

mini_auto_prompt_64k gpt-4o-mini-transcribe · 64k · auto · mixed prompt — MER accuracy pending Ground Truth · 853 ms · verdict: unjudged

老師,I want a juice, too, please.

mini_zh_prompt_deg gpt-4o-mini-transcribe · degraded · zh · mixed prompt — MER accuracy pending Ground Truth · 692 ms · verdict: unjudged

老師,I want a juice, too, please.

mini_zh_prompt_clean gpt-4o-mini-transcribe · clean · zh · mixed prompt — MER accuracy pending Ground Truth · 713 ms · verdict: unjudged

老師,I want a just to face.

gpt_tw_en_bare_deg gpt-transcribe · degraded · zh-tw + en · bare — MER accuracy pending Ground Truth · 711 ms · verdict: unjudged

老師，I want a juice, two please.

gpt_tw_en_prompt_deg gpt-transcribe · degraded · zh-tw + en · mixed prompt — MER accuracy pending Ground Truth · 764 ms · verdict: unjudged

老師, I want a juice, two please.

gpt_tw_en_prompt_clean gpt-transcribe · clean · zh-tw + en · mixed prompt — MER accuracy pending Ground Truth · 666 ms · verdict: unjudged

老師, I want a juice too, please.

gpt_tw_en_prompt_64k gpt-transcribe · 64k · zh-tw + en · mixed prompt — MER accuracy pending Ground Truth · 765 ms · verdict: unjudged

老師, I want a juice too, please.

gpt_tw_only_prompt_deg gpt-transcribe · degraded · zh-tw · mixed prompt — MER accuracy pending Ground Truth · 645 ms · verdict: unjudged

老師, I want a juice too, please.

Winning configuration

Record the variant and rationale to hand to qa-rails after reviewing this Take.