Take audit: FT-TIskFn-nQhqupL5JMh
2026-08-22T09:29:41.710Z — ray
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 · 470 ms · verdict: unjudged
Can I have a juice, please?謝謝老師
mini_auto_bare_deg gpt-4o-mini-transcribe · degraded · auto · bare
— MER accuracy pending Ground Truth · 551 ms · verdict: unjudged
Can I have a juice, please? 謝謝老師
mini_auto_bare_clean gpt-4o-mini-transcribe · clean · auto · bare
— MER accuracy pending Ground Truth · 710 ms · verdict: unjudged
Can I have a juice, please? 謝謝老師。
mini_auto_prompt_deg gpt-4o-mini-transcribe · degraded · auto · mixed prompt
— MER accuracy pending Ground Truth · 492 ms · verdict: unjudged
Can I have a juice, please? 謝謝老師.
mini_auto_prompt_clean gpt-4o-mini-transcribe · clean · auto · mixed prompt
— MER accuracy pending Ground Truth · 619 ms · verdict: unjudged
Can I have a juice, please? 謝謝老師.
mini_auto_prompt_64k gpt-4o-mini-transcribe · 64k · auto · mixed prompt
— MER accuracy pending Ground Truth · 562 ms · verdict: unjudged
Can I have a juice, please? 謝謝老師
mini_zh_prompt_deg gpt-4o-mini-transcribe · degraded · zh · mixed prompt
— MER accuracy pending Ground Truth · 516 ms · verdict: unjudged
Can I have a juice, please? 謝謝老師
mini_zh_prompt_clean gpt-4o-mini-transcribe · clean · zh · mixed prompt
— MER accuracy pending Ground Truth · 568 ms · verdict: unjudged
Can I have a juice, please? 謝謝老師
gpt_tw_en_bare_deg gpt-transcribe · degraded · zh-tw + en · bare
— MER accuracy pending Ground Truth · 630 ms · verdict: unjudged
Can I have a juice, please? 謝謝老師。
gpt_tw_en_prompt_deg gpt-transcribe · degraded · zh-tw + en · mixed prompt
— MER accuracy pending Ground Truth · 599 ms · verdict: unjudged
Can I have a juice, please? 謝謝老師。
gpt_tw_en_prompt_clean gpt-transcribe · clean · zh-tw + en · mixed prompt
— MER accuracy pending Ground Truth · 766 ms · verdict: unjudged
Can I have a juice, please? 謝謝老師。
gpt_tw_en_prompt_64k gpt-transcribe · 64k · zh-tw + en · mixed prompt
— MER accuracy pending Ground Truth · 779 ms · verdict: unjudged
Can I have a juice, please? 謝謝老師。
gpt_tw_only_prompt_deg gpt-transcribe · degraded · zh-tw · mixed prompt
— MER accuracy pending Ground Truth · 617 ms · verdict: unjudged
Can I have a juice, please? 謝謝老師。
Winning configuration
Record the variant and rationale to hand to qa-rails after reviewing this Take.