How to Re-transcribe in Reduct

Re-transcribe in Reduct

When Reduct transcribes your file, it uses the first few seconds of audio to get a feel for the recording. It picks up on the speaker's voice, accent, and background noise.

If the audio changes suddenly (loud sounds, overlapping speakers, intermittent background noise), the transcription model may struggle to recalibrate, and the accuracy may drop for that section.

Re-transcribe lets you fix just that part by sending it through transcription again, without reprocessing the whole file.

Step 1: Select the part of the transcript that you’d like to Re-transcribe.

It’s important to select the segment in the transcript where the sound environment stays consistent (one speaker or one background noise profile).

Good selections:

One person speaking without interruption.
The gray boxes in Reduct with a muffled voice.
A section with the same sustained background noise throughout (e.g., a consistent hum or street noise, not a noise that cuts in and out)

Avoid selections that:

Span multiple speaker changes or a dramatic shift in background noise
Are too short (a few words rarely give the model enough to recalibrate)

Step 2: Click ‘See More’, then click Re-transcribe.

After you click and drag to select the section in the transcript. A menu will appear on the right of your selection. Click See More, then click Re-transcribe. Reduct will reprocess just that segment and update the transcript automatically.

Re-transcribe works on any section of your transcript. If you have multiple problem spots, go through each one individually for the best results.