Accurate subtitles in more than 90 languages, generated in your browser. Style them and export an MP4 with the captions burned in, or SRT or VTT. Your video never leaves your device.
Keep this page open. Switching apps pauses it until you're back.
Free, no watermark, no account. Works offline once loaded.
Adding captions is easy and only takes a few minutes. There is nothing to install and no sign up required. The captioning tool works in your browser, on a computer or a phone.
Drop a file on the page or select it from your file browser on computer or phone. The tool accepts MP4, MOV, WebM and MKV of any length. The tool can transcribe in over 90 different languages.
Once the file is selected, the captions will start processing. Speech is turned into editable text right in your browser, with every word timed precisely. The first time you use the tool, the speech model is downloaded which may take several moments. Subsequent videos will be much quicker.
No speech recognition tool is perfect, and occasionally mistakes happen that need fixing. To do that, simply click any line in the transcript to correct it. “Problem words” such as swear words are highlighted in yellow. When you're done editing, choose the pace and style for your captions to make them pop!
Export your finished work as an MP4 with the captions burned into the picture, ready for Reels, TikTok or Shorts. Or take an SRT or VTT for YouTube and your editor, or the transcript as a plain text file.
Increasingly people are watching video content online with the sound off. This is particularly true for short form video content, for example TikTok and YT Shorts. Published figures for Facebook and LinkedIn have long put the share of views that happen muted at around 80%. Without captions, most of your audience never hears a word of it.
Captions make your video accessible. Deaf and hard-of-hearing viewers need them to follow what is being discussed in the video. Furthermore, for broadcast, publicly funded programs, and training material it's now a legal requirement in many countries. Schools and universities can caption lessons without the video leaving the classroom: see our privacy-first video tools for education.
They help people reading in a second language. Speech that's hard to follow by ear is easy to follow on the page, so subtitles reach viewers who would otherwise scroll past.
They can help grab attention, when done well. When subtitles are styled well and fit with the on-screen action, they can help to grab and keep the audience's attention for longer.
YouTube reads your caption text. It can surface a video for the words spoken in it, and in a study by Discovery Digital Networks, captioned videos took about 13% more views in their first two weeks. Correcting names and jargon before you export makes that worth more.
It remains yours. Oh So Video is a free subtitle generator that runs entirely in your browser: no watermark, no sign-up, and unlike most online caption tools, it never sees your footage. Any size of video works, up to three hours long for desktop computers and an hour on a phone (due to processing and memory constraints).
Speech recognition, editing and exporting all happens within your browser. Your video and its transcript are never uploaded anywhere. No cookies, no ad trackers, no account. You'll notice that we don't run cookie banners on the website — because we don't have to.
Names and jargon are where speech recognition can make mistakes. Click any line in the transcript editor to correct it: words that are often misheard are flagged, and every edit keeps its timing.
Adjust the pace of the captions to match the style of your production and the preferences of your viewers. Slide from punchy one-to-three-word captions for Reels, TikTok and Shorts to classic two-line subtitles for YouTube and film, where the tool keeps it all within broadcast reading limits.
Five caption styles, with an animation if you want one, a size, a position and any highlight colour, with the text on it kept readable for you. Export an MP4 with them drawn into the picture, exactly as the preview shows — no watermark, made on your device.
Spanish, French, German, Portuguese, Italian, Russian, Arabic, Hindi, Turkish and more. The tool automatically detects which language is spoken from the first half-minute, or you can choose it yourself.
Every word is timed on its own so that captions appear as the words are spoken, not a beat late. Silent stretches are skipped, so nothing gets invented in the gaps.
YouTube. In YouTube Studio, open Subtitles, choose your video and add a language, then add subtitles from a file. Say that it includes timing, and pick your .srt. Viewers can choose to turn them on and off with the CC button, and YouTube can read the words in your captions when it works out what the video is about.
Premiere Pro, DaVinci Resolve, Final Cut Pro and CapCut. All four take an SRT as a caption track. This gives you more flexibility over the style and positioning of your captions. Restyle it, move it, and burn it into the picture on export if you want it on screen in the final video.
Facebook, LinkedIn, Vimeo. Each takes a subtitle file alongside the video when you post it, so the captions can be switched on and off rather than being part of the picture. Vimeo and web players want the VTT; the rest take the SRT.
Any video player on your own machine. VLC, QuickTime and the rest look for a subtitle file with the same name as the video, which is exactly what you get here. Export it next to your video and it loads itself.
Of course. There is no account to sign up for, no trial to slip you up, and no watermark on anything you export. Oh, and no limit on how many videos you caption.
No. Your browser reads the file, pulls out the audio and transcribes it with a speech model that runs on your own device. Nothing is ever uploaded to a server, the video and its transcript are never sent to us or anyone else. The only things your browser downloads are the website page code and the open source speech model.
Captions are for people who can't hear the sound, so they carry who is speaking and what else can be heard. Subtitles assume you can hear but may not follow the language, so they carry the speech alone. In practice the words are used interchangeably, and one file works for both: Oh So Video makes SRT and VTT, which platforms like YouTube show as closed captions.
Ultimately it depends on the quality of sound and who is speaking: clear speech in a quiet room transcribes very well, whilst background noise or a strong accent, less so. Transcription uses OpenAI's Whisper speech model. Names, brands and jargon are where it slips, so every line is editable. In English, words that are often misheard are flagged in yellow, and your corrections keep their timing. The file is as accurate as you make it before you export.
Absolutely. When you're ready, choose Export, then .txt: the transcript as it reads, one line per caption line, with no times and no numbering. These are the words as you've edited them, ready for notes, a blog post, show notes or subtitles that can be timed manually if you want to do that.
Currently the tool accepts MP4, MOV, WebM and MKV. These can be videos from phones, cameras and screen recorders. Only the sound is needed, so a video your browser can't show still transcribes — you hear it and caption it, with a still frame in place of the picture.
Nope. It all happens in this browser tab. There is no app, no extension, no plug-in and no account needed. Once the page and the speech model have loaded, it even works without an internet connection.
SRT (SubRip) works almost everywhere: YouTube, Facebook, LinkedIn, Premiere Pro, DaVinci Resolve, Final Cut Pro and CapCut. VTT (WebVTT) is the format for web players, such as HTML5 video and Vimeo. Both carry exactly the same text and timings. There is also .txt for the words alone, and .mp4 for the video with the captions burned in.
Export the .srt file. In YouTube Studio, open Subtitles, choose your video and add a language. Choose to add subtitles from a file, say that it includes timing, and select your .srt. Viewers can then turn the captions on with the CC button, and YouTube can read the words in them.
Yes, all four import SRT files. Bring the .srt into your project and it becomes a caption track you can restyle, and burn into the picture when you export if you want open captions.
Yes. SRT and VTT are plain text, so any text editor opens them, and the timings are readable lines you can nudge. It's usually easier to fix the words here first, where every edit keeps its timing and you can watch the preview.
Yes, in more than 90 languages, including Spanish, French, German, Portuguese, Italian, Russian, Arabic, Hindi and Turkish. It works out the language from the first half-minute of speech, or you can choose it under the drop target. The captions are in the language spoken: it doesn't translate them. Accuracy is best in widely spoken languages, and English is the most accurate of all. Chinese, Japanese and Thai aren't supported yet.
There is currently no limit to the size of a file. With length, it's three hours long on a computer and an hour on a phone. Length is the limit rather than size, because the transcribing happens on your device. You should expect approximately 10 minutes of work per hour of video on a computer, and longer on a phone, which has less memory and a slower processor. Long videos are best done on a computer. Ultimately it depends on the processing power of your device. Older devices will of course take longer to process the video transcript.
To transcribe on your device rather than on a server, your browser needs the speech model. This is approx 340 MB on a computer, which gets larger, more accurate models, and up to 170 MB on a phone. It's fetched once and then kept by your browser, so after that new videos start straight away. Keep in mind that if you are using the tool on a phone, that it is best to be connected to WiFi, to avoid excessive data usage through your carrier.
It does, providing you are running an up-to-date browser on iPhone or Android. Transcribing takes longer on a phone, so keep the page open while it works; switching apps pauses it until you come back. Your screen stays on while it runs. Accuracy of the transcription is also reduced on a phone, due to the tool using the more lightweight speech recognition model.
Your work is saved in the browser. The transcripts of your five most recent videos, with your edits, are saved in your browser, on your device, for 14 days. Choose one of those videos again and you carry on where you left off, without waiting for it to transcribe again.
Yes. Choose a look under Style: Classic, Minimal, Bold, Highlight (the word being said changes colour) or Pill, with an animation if you like (Bounce, Pop, Fade or Reveal), a size, a position and a highlight colour: one of six, or any colour you pick. On the Pill style the text turns black or white by itself, whichever stands out more against your colour. Then choose Export as a .mp4. This will create a video with the captions burned into the picture, exactly as the preview shows them, made on your device. Or export an SRT or VTT file for YouTube or your editor.