By ShinobiTools Team · Last updated: July 2026
Recorded a call? Upload the Zoom, Teams or Google Meet recording and get an accurate transcript back: searchable notes, action items and a written record of what was said. No account, no cap.
Transcribe audio now →Meeting recordings are hard to search and painful to re-watch. A transcript gives you a written record you can skim, search and quote: decisions, owners and deadlines all in text. Pair it with our meeting minutes from audio page for turning that transcript into clean minutes.
Zoom, Teams and Meet all let you save a local recording (usually MP4 audio+video, or M4A audio). Upload whichever you have; ScribeGrab only needs the audio and will pull it from a video file automatically.
Meetings are sensitive, so your file is deleted the moment transcription finishes and results are wiped within 45 minutes. Nothing is kept, shared or used to train anything. Details on the privacy page.
ScribeGrab works on a file, and every meeting platform saves one. The only real question is where it went, and each of the big three hides it somewhere different.
Documents/Zoom/ in a folder named after the meeting date and title. The file you want is audio_only.m4a if it's there (smallest and perfectly adequate), otherwise zoom_0.mp4.Two practical notes. An audio-only download is always the better choice when the platform offers one: it's a fraction of the size, uploads far quicker and transcribes identically, because the video track is discarded anyway. And if the platform gives you a transcript of its own, it's usually worth transcribing again regardless: built-in meeting transcripts are tuned for speed over accuracy and tend to fall apart exactly where the meeting got interesting.
A one-hour meeting transcribed as a single block of text is technically complete and practically useless. You can't tell who committed to what, and you can't quote anybody. ScribeGrab separates the voices automatically and labels them Speaker 1, Speaker 2 and so on, up to eight distinct speakers, with nothing to switch on.
That changes what you can do with the result. You can search the transcript for one person's contributions, see at a glance who dominated the hour, and lift a decision with the name attached. The first thing worth doing is a find-and-replace: swap Speaker 1 for the actual name once and the whole document becomes readable. Work out who is who from the first thirty seconds, where people almost always introduce themselves or are greeted by name.
Where it's honest to expect trouble: labels are AI-estimated from voice characteristics, not from the platform's participant list, so two speakers with similar voices on similar microphones can occasionally be merged, and crosstalk (two people talking over each other) tends to get assigned to whoever is louder. If a recording has only one voice throughout, no labels are added, which is correct rather than a failure. And a call with more than eight distinct voices falls back to plain unlabelled text, so for a large all-hands expect a transcript rather than a cast list. The click-to-play transcript on the results page is the fix for all of these: click any line and it plays that moment of the audio, so verifying a label or a quote takes a second instead of a scrub through an hour.
Yes. Save the local recording from any of them and upload it. ScribeGrab reads both the audio-only and the video recording.
It produces one accurate transcript of everything said. It does not label speaker names, so add those yourself if you need them.
No. Files up to 90 minutes and 2 GB fit in a single upload, with no daily cap.
Yes. The file is deleted right after processing; results wiped within 45 minutes.