For schools, colleges, universities and public services
Caption, transcribe, compress and trim classroom video for free, in the browser. Nothing is uploaded, nobody needs an account, and there's no cookie banner, because there's nothing to ask about.
The problem
Every school runs on video now. Lessons recorded for students who were off sick. Presentations, drama and PE assessments. Interviews for a history project. Assemblies, open days and the clip of the choir that every parent wants a copy of. Most of it shows children, and much of it shows them by name.
And most of it gets edited the quickest way someone can find: by searching for a free tool, dropping the file in and waiting. That quick fix is where the risk is, because of how nearly every free online video tool works.
A typical online video editor, compressor or caption generator doesn't do the work in your browser. It takes a copy of your file, sends it over the internet to a server, does the work there and sends a result back. For those minutes, and often for much longer, a complete copy of the video sits on a computer the school doesn't own, in a country it may not know, run by a company it has never checked.
What happens next is in the small print, if it's anywhere. Some services delete files after an hour; some keep them for weeks; some keep them until an account is closed. Some terms grant the service a licence to use what you send "to improve our products", which can include training AI models. Many pass files to other companies for storage, transcription or analytics. A free service has to be paid for somehow, and the data that passes through it is the obvious place to look.
A school handles the personal data of children, and the law treats that as a special responsibility: the UK's data protection regulator, the ICO, explains why children merit specific protection. A video of a pupil who can be identified is personal data in its own right: a face, a voice, often a name and a uniform that says where they go every day. Data protection law in the UK, Europe and elsewhere expects a school to know where that data goes, why, for how long and under what agreement. A teacher who drops a clip into an unknown website to shrink it for email has, without meaning to, handed it to a supplier nobody approved.
The usual answer is a list of approved tools, a contract with each supplier and a data protection impact assessment for every one. The Department for Education's guidance on procuring educational technology asks schools to check where a supplier holds pupils' data, whether it leaves the UK, and what the contract says about deleting it. That's the right process, and it's slow, which is exactly why people reach for the quick free tool instead. The better answer is a tool that leaves nothing to assess, because the video never goes to a supplier in the first place.
The solution
Oh So Video turns that model around. Instead of sending your video to the tools, the page brings the tools to your video. When you open a tool, your browser downloads the program that does the work, the same way it downloads any web page. When you choose a video, the browser reads it straight from the device's own storage and the work happens there, on the device's own processor and graphics chip. When it's finished, the result is saved back to the same device.
At no point does the video travel anywhere. Not the file, not a single frame, not its sound, not its name. The website can't see it, because it's never sent to the website.
Any browser, any device. Nothing to install, no account, no sign-in.
The browser reads it from the device. It isn't uploaded or copied anywhere.
Its own processor and graphics chip, in the background, so the page stays responsive.
Straight to the same device, with no watermark, ready to share as you choose.
The tools
Each does one job well, in the browser, with the same privacy. These are the ones schools reach for most.
Subtitle lessons and assemblies for deaf and hard-of-hearing students, and for anyone watching with the sound off.
Subtitles in 90+ languages: burn them into the video, or export SRT.
Open captions generatorTurn a recorded lecture, interview or oral exam into text to read, search and quote.
Video or audio into editable text, in 90+ languages. Plain or timestamped.
Open transcript generatorGet a project video under the size your learning platform or email will take.
Make a video smaller, or made for email, Discord, WhatsApp or social.
Open compress videoCut a clip to the exact moment for a lesson, without re-encoding it.
Cut the start and end to the exact frame, in the original quality.
Open trim videoRecord a flipped-classroom explainer or a welcome message from a script.
Record yourself reading from a script that scrolls by the camera.
Open record with teleprompterMake an iPhone or iPad video play on the school's Windows PCs and projectors.
Turn an iPhone or camera MOV into an MP4 that plays everywhere.
Open convert mov to mp4Frame out what shouldn't be seen, or reshape a clip for the school's social accounts.
Crop to 9:16, 4:5, 1:1 or 16:9 for TikTok, Instagram and YouTube, or any shape.
Open crop videoTake the sound off before a video is shared, voices and background music included.
Take the sound off any video, keeping the picture exactly as it is.
Open remove audio from videoKeep the audio of a talk or performance as an MP3 for a podcast or listening task.
Take the sound from any video as an MP3, at the quality you choose.
Open video to mp3Send a performance or a project home to parents in a single email.
Fit one video, or several together, into a single email.
Open compress video for emailShrink a whole class's set of project videos in one go.
Compress several videos at once, and save them all together.
Open batch compress videosWho it's for
Caption, trim and compress lesson videos between classes, with nothing to install and no account to make.
Finish a project video on a school laptop or their own phone, free, with no watermark on the result.
Subtitles in more than 90 languages, and transcripts for students who need the words as well as the picture.
Nothing to host, no supplier holding student video, and a short list of the hosts the page contacts.
Crop for 9:16, compress for social, burn in captions: open-day and event video, ready to post.
Work on interviews, meetings and casework recordings that must never sit on a stranger's server.
In detail
"We take your privacy seriously" is easy to write. Here is exactly what happens instead, in enough detail for an IT manager or a data protection officer to check.
Modern browsers can do far more than show pages. They can read a file the user chooses, decode and encode video with the device's own hardware (an interface called WebCodecs), and run substantial programs, including AI models, at close to native speed through WebAssembly. Oh So Video is built entirely on those abilities. Reading and writing video files is done by Mediabunny, an open-source library. Captions and transcripts come from Whisper, OpenAI's open speech recognition model, running inside the browser through ONNX Runtime. The heavy lifting runs in Web Workers, separate threads in the page, so the page stays responsive while a video is processed.
None of this needs a server to look at the video. There is no upload step that's hidden, sped up or encrypted; there's simply no upload. The site has no back end that could receive a video even if the page tried to send one: it's a set of static files.
A web page has to fetch its own files, and a few small things are counted so we know the tools are working. This is the complete list of hosts a tool page contacts, and what each one gets:
| Host | What it's for | Anything from your video? |
|---|---|---|
ohsovideo.com | The page itself, its font, icons and the code that runs the tools. | No |
cdn.ohsovideo.com | Our own CDN: the speech models for captions and transcripts, and the engine that runs them. Fetched once; the browser keeps them. | No |
plausible.io | A cookieless count of visits, and the name of a tool's step ("Trim started"). | No |
cloudflareinsights.com | A second cookieless visit count, being retired. | No |
eu.i.posthog.com | Only if a tool fails, or someone sends the optional feedback card: what went wrong and the kind of file ("MOV, 100 MB to 1 GB"). | No: never the video, its name or anything in it |
For the tools to work, a web filter needs to allow only the first two. The other three can be blocked and nothing breaks. Nothing loads from Google, from social networks or from advertising networks.
The speech models are the one large download: between about 80 and 250 MB, depending on the language and the device, the first time someone makes captions or a transcript. The browser keeps them, so the next video starts straight away. On a busy school network that's worth knowing; it's the model, not the video, and it only comes in.
A few things are kept in the browser's own storage, on the device, so the tools are pleasant to use: the light or dark theme, a tool's last setting, and the transcripts of your five most recent captioned videos for up to 14 days, so a refresh doesn't lose an afternoon's editing. None of it is sent anywhere. "Clear saved work" on the captions page deletes the transcripts, and clearing the browser's data for the site removes everything. On a shared classroom computer, that's worth doing at the end of a lesson.
You don't have to take our word for any of this. Open the browser's developer tools, choose the Network tab and process a video: you'll see the page's own files and the requests in the table above, and nothing the size of your video going out. The privacy policy lists everything the site sends and keeps, in plain words.
The libraries the tools are built on are open source under permissive licences: Mediabunny (MPL-2.0), ONNX Runtime (MIT) and the Whisper models (MIT). The MP3 encoder, LAME, is used unmodified under the LGPL and credited in the privacy policy. There's no copy of FFmpeg's GPL build anywhere in the site.
ohsovideo.com and cdn.ohsovideo.com.
Yes, and for everyone else. There's no education plan to sign up for, no per-seat price, no watermark on what you export and no trial that runs out. It costs us almost nothing to run because your own device does the work, so there's nothing to charge for.
No. Open the page and choose a video. With no accounts there's no list of names, emails or passwords for us to hold, lose or be asked for, and nobody has to agree to anything before they start.
We can't certify that for your school, but we can tell you the facts your data protection lead needs: the video is read, processed and saved on the device in front of you and is never sent to us or anyone else. The site sets no cookies and keeps no identifier. Visits are counted with Plausible, which stores no personal data, and the occasional error report or optional feedback goes to PostHog's EU cloud with no identifier and without the file. The privacy policy has the detail.
ohsovideo.com for the pages and tools, and cdn.ohsovideo.com, which serves the speech models the captions and transcript tools download the first time. Nothing else is needed for the tools to work; the analytics hosts can be blocked without breaking anything.
Yes. Every tool runs in a current browser: Chrome, including on Chromebooks, Edge, Safari on Macs, iPads and iPhones, and Firefox. Long videos are quickest on a laptop or desktop. Firefox can't convert or compress iPhone HEVC video, and says so; Chrome, Edge and Safari can.
Compressing and converting take a video of any size. Trimming and removing audio take up to 2 GB. Captions and transcripts take any size of file up to 3 hours long on a computer, or an hour on a phone. The limits are your device's rather than ours, and each page says its own before you choose a file.
That's what it's built for. A video of a child is exactly what shouldn't be handed to a service you know nothing about. Here it never leaves the school's device, so it's held by whoever holds that device, under your own policies, and nobody else.
Choose a lesson recording, a project or an assembly clip. It stays exactly where it is.