Skip to content

Use case

Captions for interviews

Interview footage is the most sensitive material most creators ever handle: sources who were promised anonymity, subjects under NDA, recordings from stories that haven't published. It's exactly the material that should never be uploaded to a captioning service and exactly the material that needs a good transcript.

The privacy case, stated plainly

OneDrop transcribes without uploading: the recording is read from your disk into the tab, and the video file never leaves your device, only the compressed audio is transcribed on the server and deleted right after processing. There is no video upload to misconfigure and no lasting service-side copy of an un-broadcast interview. For research ethics boards and source agreements alike, "the recording never left my device" is an answer you can actually give and verify, per the privacy guide.

Two-voice audio, handled honestly

Interviews are the hardest common audio: overlapping speech, laughter over answers, remote-call compression. The studio finds the speech regions, leans on the voice over the noise, and, importantly, tells you when audio genuinely covers the words instead of inventing a plausible guess. In an interview context, a flagged "couldn't hear this" is worth more than a confident wrong quote.

The transcript is the deliverable

Beyond burned-in captions for the published cut, the TXT export is the working transcript writers and researchers actually use: pull quotes with timestamps, build the paper edit, hand the text to an editor or translator. Because the transcript is word-level accurate to the timing, a quote in your notes maps straight back to the moment in the footage.

A workflow that respects embargoes

  1. Drop the interview recording into the studio, the video never uploads.
  2. Transcribe; fix names and terms (the flagged-word list speeds this up).
  3. Export TXT for the edit, SRT/VTT if the cut goes to a platform that attaches subtitles, MP4 with burned-in captions for social cuts.
  4. Clear saved data from the footer when the project closes, the only copy was yours anyway.

Limits, honestly

The free plan covers 2 minutes per video, 2 minutes of audio per day, resetting at 00:00 UTC, enough for clips and highlight cuts, not a 90-minute sit-down. Longer interviews are what the paid plans are for; Starter, Pro and Studio are live now, monthly plans with up to 30-minute videos. And the same advice as always applies: automated transcription is a first draft. For published captions, run the five-minute review routine before anything ships.