SCRIBEGRABOpen the tool →

About ScribeGrab

No account. No daily cap.Others give you three files a day and call it unlimited. Files up to 60 minutes here, as often as you like.

Upload a recording, get the words back as a text file and as subtitles, and the audio is deleted the second the transcript exists.

What this is

You drop in a recording, a graphics card in my house runs it through our speech model, and you get back a plain-text transcript plus time-coded SRT and VTT subtitles. No account, no watermark, no daily cap, no per-minute meter ticking along in the corner.

The "my house" part is literal. I am one person with one machine, and I also built StemGrab and PicReviver, which run on the same two cards. When all three are busy your file waits in a queue behind somebody's song. That is the honest reason a transcript is sometimes instant and sometimes takes a minute.

Specifications

The hard numbers, so you know before you upload. Every value here is what the server actually enforces.

FreeWith a pass
PriceFree. No account, no sign-up, no card.€5 for 30 days, or €3 for 24 hours
Maximum length60 minutes per file4 hours per file
Maximum file size1.5 GB1.5 GB
Files at once25
QueueNormalPriority when the cards are busy
Results kept for45 minutes24 hours
Speaker labelsAdded automatically on both tiers — "Speaker 1:", "Speaker 2:" in the transcript and subtitles
Renaming the speakersNoYes — replace "Speaker 1" with a real name throughout
Word (.docx) exportNoYes
Meeting minutesNoYes
Translate intoOne language at a timeUp to 5 at once
Transcribing per dayNo capNo cap
Text-to-speech per day10 minutes60 minutes
Text tools per day25,000 characters250,000 characters
Input formats25 audio and video formats: MP3, WAV, FLAC, M4A, AAC, OGG, OPUS, WMA, AIFF, AIF, MP4, MOV, MKV, WEBM, AVI, M4V, MPEG, MPG, WMV, 3GP, AMR, M4B, CAF, TS, FLV
Output formatsTXT, SRT, VTTTXT, SRT, VTT, DOCX
Transcription languagesAbout 100, detected automatically
Translation languages16: English, Spanish, French, German, Dutch, Portuguese, Italian, Polish, Turkish, Russian, Arabic, Hindi, Indonesian, Japanese, Korean and Simplified Chinese. Timestamps are kept.
Where it runsOn our own graphics cards in the Netherlands. Your audio is never sent to OpenAI, Google, AWS or any other transcription service.
Your uploadDeleted the moment processing finishes, success or failure — on both tiers.
WatermarkNone, on any output, on either tier.

The free tier is not a trial and does not run out. What the pass buys is length, speed and the extra export formats — the transcription itself is the same model either way.

Why it's free: the honest version

Otter, Rev, Sonix and the rest pay a cloud provider for every minute of audio they touch, so they have to charge you by the minute or cut you off after a few files. I do not have that bill. The cards are bought and paid for; your hour of audio costs me some electricity and nothing else.

The ads on the page cover that electricity and the domain name. There is also an optional pass — €5 for 30 days, €3 for 24 hours — which raises the limit from 60 minutes to 4 hours, adds speaker labels, Word export and meeting minutes, and puts you ahead in the queue when the cards are busy. The exact difference is in the table above. What it does not do is take anything away from the free tier: the SRT was always going to be free, it is the same file with timestamps in it, and no free limit was lowered to make the pass look better.

Privacy: your file doesn't stick around

Transcription only works if you hand me your recordings, so the rules are short and I do not bend them:

The full details are in the Privacy Policy.

What it's good at (and what it isn't)

Clear speech — interviews, lectures, podcasts, voice memos, screen recordings — comes back very accurate, in around 100 languages that it detects by itself. Three things beat it, and they beat every other system too: a room full of background noise, people talking over each other, and a very strong accent on a bad microphone. I wrote an honest guide about how accurate automatic transcription really is rather than quote a percentage at you. Music does not transcribe at all; for splitting songs into stems, use StemGrab.

Questions or problems?

Found a bug, got a strange result, or just want to say something? The contact form lands in my inbox, not a ticket queue.

Try the tool →

About the name

About the byline. I write as Sam Ridder. It is a pen name: transcription is a thing people hand private recordings to, and I would rather they judged the setup I have described here than a name they can look up. Every factual claim on this page is accurate.