How to apply this guide
Preserve question boundaries while checking whether silence or weak speech caused omissions.
Start with
Listen near the start and end of each answer. Mark whether the gap is actual silence, a quiet response, or background noise. The technique differs if information was spoken very softly.
Check before you finish
Write the question or an identifier above each result. If a passage remains unclear, flag it rather than inferring the answer from the question.
Evidence to keep: Keep the target field open and verify that the complete reviewed result reached the intended destination. This page’s key limit is: Preserve question boundaries while checking whether silence or weak speech caused omissions.
Start here
A recording with long gaps can look empty to a recognizer, even though individual answers are audible. Split or trim a copy at known boundaries instead of speeding up the entire file.
Review the audio first
Listen near the start and end of each answer. Mark whether the gap is actual silence, a quiet response, or background noise. The technique differs if information was spoken very softly.
Use one answer per test
Export one clearly audible response for a first import. If it transcribes correctly on its own, split the rest at good boundaries. Retain the original file and offsets so you can review later.
Keep the question with the answer
Write the question or an identifier above each result. If a passage remains unclear, flag it rather than inferring the answer from the question.




