← All runs

Evaluation run

first live run

6 July 2026, 10:24 · eval-engine · f84da06 · 87 cases

Did it hear the words right?

This run graded typed text only, so there is nothing here to have misheard.

Did it clean up correctly and safely?

79%
Light, ships on by default
70/89 cases passed · 346 ms typical
72%
Polish, experimental
64/89 cases passed · 403 ms typical
When a case fails, whose fault is it?
0 the microphone misheard the words  ·  44 the cleanup itself slipped

Every case

Each case shows the instant on-device cleanup, then what each mode did to it. Added words are marked, removed words are struck. For a spoken case, Heard is what the microphone transcribed and Was said is the reference it is scored against.

87 cases match

num-one-of-themTypedclean
Instantone of them is already broken
LightOne of them is already broken. 230 ms
PolishOne of them is already broken. 280 ms
num-one-on-oneTypedneeds review
Instantlet's do a one on one tomorrow
LightLet's do a one on one one-on-one tomorrow. 352 ms
PolishLet's do a one on one one-on-one tomorrow. 364 ms
Why?the result is missing “one on one”
num-one-secTypedclean
Instanti'll be ready in one sec
LightI'll be ready in one sec. 319 ms
PolishI'll be ready in one sec. 333 ms
num-percentTypedclean
Instantrevenue grew 25% last quarter
LightRevenue grew 25% last quarter. 276 ms
PolishRevenue grew 25% last quarter. 349 ms
num-phoneTypedclean
Instantcall me at five five five one two three four
LightCall me at five five five one two three four 555-1234. 458 ms
PolishCall me at five five five one two three four 555-1234. 485 ms
num-rangeTypedclean
Instantit should take 2 to 3 days
LightIt should take 2 to 3 days. 307 ms
PolishIt should take 2 to 3 days. 386 ms
num-roomTypedneeds review
Instantthe meeting is in room 2:05
LightThe meeting is in room 2:05. 334 ms
PolishThe meeting is in room 2:05. 405 ms
Why?the result is missing “205”
num-room-214Typedneeds review
Instantthe interview is in room 2:14
LightThe interview is in room 2:14. 446 ms
PolishThe interview is in room 2:14. 426 ms
Why?the result is missing “214”
num-selfcorrectTypedneeds review
Instantthe total comes to 50 no wait $60
LightThe total comes to 50 no wait $60. 273 ms
PolishThe total comes to 50 $60. No wait, $60 the total comes to $50. 650 ms
Why polish?the result still contains “$50” and “no wait”, which this case forbids
num-timeTypedclean
Instantlet's meet at 4:30 this afternoon
LightLet's meet at 4:30 this afternoon. 347 ms
PolishLet's meet at 4:30 this afternoon. 444 ms