Take audit: KAaNebuhFrgpnokz8ekRt
2026-08-22T09:21:45.206Z — 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 · 730 ms · verdict: unjudged
老師,can you help me? 我需要幫助,幫忙
mini_auto_bare_deg gpt-4o-mini-transcribe · degraded · auto · bare
— MER accuracy pending Ground Truth · 859 ms · verdict: unjudged
老師,Can you help me?我需要幫助,幫忙
mini_auto_bare_clean gpt-4o-mini-transcribe · clean · auto · bare
— MER accuracy pending Ground Truth · 803 ms · verdict: unjudged
老師,Can you help me? 我需要幫助、幫忙
mini_auto_prompt_deg gpt-4o-mini-transcribe · degraded · auto · mixed prompt
— MER accuracy pending Ground Truth · 730 ms · verdict: unjudged
老師,can you help me? 我需要幫助,幫忙
mini_auto_prompt_clean gpt-4o-mini-transcribe · clean · auto · mixed prompt
— MER accuracy pending Ground Truth · 731 ms · verdict: unjudged
老師,can you help me? 我需要幫助,幫忙.
mini_auto_prompt_64k gpt-4o-mini-transcribe · 64k · auto · mixed prompt
— MER accuracy pending Ground Truth · 879 ms · verdict: unjudged
老師,can you help me? 我需要幫助,幫忙
mini_zh_prompt_deg gpt-4o-mini-transcribe · degraded · zh · mixed prompt
— MER accuracy pending Ground Truth · 731 ms · verdict: unjudged
老師,Can you help me? 我需要幫助,幫忙。
mini_zh_prompt_clean gpt-4o-mini-transcribe · clean · zh · mixed prompt
— MER accuracy pending Ground Truth · 701 ms · verdict: unjudged
老師,can you help me? 我需要幫助,幫忙
gpt_tw_en_bare_deg gpt-transcribe · degraded · zh-tw + en · bare
— MER accuracy pending Ground Truth · 675 ms · verdict: unjudged
老師，can you help me? 我需要幫助，幫忙。
gpt_tw_en_prompt_deg gpt-transcribe · degraded · zh-tw + en · mixed prompt
— MER accuracy pending Ground Truth · 809 ms · verdict: unjudged
老師，can you help me? 我需要幫助，幫忙。
gpt_tw_en_prompt_clean gpt-transcribe · clean · zh-tw + en · mixed prompt
— MER accuracy pending Ground Truth · 1063 ms · verdict: unjudged
老師，can you help me? 我需要幫助，幫忙。
gpt_tw_en_prompt_64k gpt-transcribe · 64k · zh-tw + en · mixed prompt
— MER accuracy pending Ground Truth · 664 ms · verdict: unjudged
老師，can you help me? 我需要幫助，幫忙。
gpt_tw_only_prompt_deg gpt-transcribe · degraded · zh-tw · mixed prompt
— MER accuracy pending Ground Truth · 971 ms · verdict: unjudged
老師，can you help me? 我需要幫助，幫忙。
Winning configuration
Record the variant and rationale to hand to qa-rails after reviewing this Take.