Video unavailable. Open the source below.
NewJev ASR Reading Review
Pairs streaming speech recognition with typed reading-error checks.
Added to Jevfast
How it uses Jev
Streaming ASR supplies recognized text with a stable prefix, which the project compares with the source passage. Jev answers bounded text-review questions: whether a word matches the original, which error category applies, and whether the passage meaning drifted. The post does not specify the exact state fields, error options, score criteria, or how results are shown to the reader.
Why it is interesting
Stable prefixes let downstream judgments operate on committed recognition text instead of repeatedly revising decisions as the transcript changes.