TryScriptorium fact sheet

TryScriptorium is a web app that turns speech in audio and video files into editable, timestamped text. It supports 33 languages plus automatic detection, including Arabic, and exports TXT, SRT, VTT, Markdown, JSON and CSV. One file at a time, up to 15 minutes and 25 MB, 3 tries per day per network address, no sign-up and no card.

Last updated: 24 October 2026

This is a fact sheet: short statements that can be quoted on their own. Limits and prices are read live from the running service, so they match what the product enforces. Nothing here is an estimate, a rating or a user count.

What it is

• A hosted web application at tryscriptorium.com that transcribes uploaded or recorded audio and video.
• Speech recognition is faster-whisper, an optimised runtime for OpenAI's open Whisper models, running on our own server. Recordings are not sent to a third-party transcription API.
• Every line has a timestamp. Speaker labels (up to 8 speakers), inline editing and full-text search across transcripts are included.
• It can also translate speech in another language into English text.
• The same software can be self-hosted for recordings that must not leave your own infrastructure.

Who it is for

People who have a recording and need the text: meetings, interviews, lectures, podcasts, voice notes and videos. It works on files you already have. It does not join live calls and does not produce real-time captions.

Free and paid

• No account: One file at a time, up to 15 minutes and 25 MB, 3 tries per day per network address, no sign-up and no card.
• Free account: keeps transcripts in a library and uses the account limits below.
• Paid plans: monthly subscriptions that add transcription minutes. Prices below are read from the plan catalogue.

Details and the current allowance are on the pricing page.

Languages

33 languages plus automatic detection: English, Arabic, Spanish, French, German, Portuguese, Italian, Dutch, Turkish, Russian, Ukrainian, Polish, Hindi, Urdu, Bengali, Chinese, Japanese, Korean, Vietnamese, Thai, Indonesian, Malay, Tagalog, Persian, Hebrew, Swedish, Danish, Norwegian, Finnish, Greek, Czech, Romanian, Hungarian.

Arabic has its own pages: Arabic audio and video to text and تفريغ الصوت العربي.

File formats

• Upload, video: MP4, MOV, MKV, WebM, AVI, M4V, WMV, FLV, MPEG, 3GP, TS.
• Upload, audio: MP3, WAV, M4A, AAC, FLAC, OGG, Opus, WMA, AIFF, AMR, CAF.
• Export: TXT, TXT with timestamps, SRT, VTT, Markdown, JSON (word timings), CSV.

Limits

• No account: One file at a time, up to 15 minutes and 25 MB, 3 tries per day per network address, no sign-up and no card. The preview uses the "small" model.
• Signed in: By default, files up to 4,096 MB per upload and no fixed length limit per file; the operator can change these, and monthly transcription minutes depend on the plan.
• Queue: transcription runs on one shared worker, so long files wait their turn. Pro+ jobs go ahead of normal jobs that are still waiting.
• Not supported today: Pasting a YouTube or other web link — upload the file itself instead; Burned-in captions rendered into the video file; A public API.

Privacy and data retention

• Uploads go over HTTPS to the TryScriptorium server and are processed there.
• No-account preview files, and the temporary account that owns them, are deleted automatically after 24 hours.
• With an account, files and transcripts stay until you delete them. Accounts are isolated from each other.
• Uploads are not used to train models and are not sold or shared with third parties.
• For privileged, medical or contractually restricted recordings, self-host instead of using the hosted site.

Full wording: privacy policy and security.

Accuracy

No accuracy percentage is published for your audio, because it depends on the recording. Treat a transcript as a draft and check names, numbers and quotes. For measured speed and word error rate on public test clips, with the method and hardware stated, see Transcription speed and accuracy, measured.

How it compares, honestly

• Versus meeting-bot tools that join Zoom, Meet or Teams: TryScriptorium never joins a call. You upload the recording afterwards. There are no live captions.
• Versus YouTube's own transcript: YouTube only offers it for videos that have captions. TryScriptorium transcribes any file you have, including private recordings, and lets you edit and export it. It cannot fetch a YouTube link; you upload the file.
• Versus human transcription services: machine transcription is faster and cheaper but needs checking. For certified legal or medical transcripts, use a human service.
• Versus running Whisper yourself: the model family is the same. The hosted site adds an editor, search, speaker labels and export so there is nothing to install. The software can also be self-hosted.

Where to verify

• Pricing and limits
• llms.txt and llms-full.txt (machine-readable summary)
• Measured speed and accuracy
• Contact