How to Add Subtitles to Any Video for Free (Without Uploading It)
Captions stopped being optional. Most social video is watched muted, captioned clips hold attention longer, and subtitles are the single biggest accessibility win available to a small creator. The catch has always been that captioning tools want your video on their servers, your email address, and eleven dollars a month.
What actually happens when you caption a video
- 1The audio is separated from the video — captioning never needs the picture.
- 2A speech-recognition model turns that audio into words with timestamps.
- 3Words are grouped into caption lines short enough to read in the time available.
- 4You get either a sidecar subtitle file (.srt or .vtt) or captions burned into the pixels.
That distinction in the last step matters more than people expect. A sidecar SRT is a tiny text file you upload alongside your video — YouTube, Premiere, and CapCut all read it, viewers can toggle it off, and search engines can index it. Burned-in captions are permanent pixels, which is what you want for Instagram, TikTok, or anywhere subtitle files are not supported.
How the popular options compare
Kapwing, Veed, and Descript are strong products with real editing suites behind them — if you are cutting a podcast into clips every week, they are worth the subscription. CapCut is free and excellent on mobile. The trade-offs are consistent, though: your footage is uploaded, free tiers cap minutes or add watermarks, and you need an account.
| Upload-based caption apps | Browser-local captioning | |
|---|---|---|
| Your video goes | To their servers | Nowhere — stays on your device |
| Speed | Limited by your upload speed | Limited by your CPU |
| Free tier | Minute caps or a watermark | Unlimited, no watermark |
| Works offline | No | Yes, after the first load |
| Editing suite | Full timeline editors | Captions only |
| Best for | Weekly production workflows | One-off clips and private footage |
Use a full suite when captioning is one step in a bigger edit. Use something local when the video is sensitive, when it is a quick one-off, or when your upload speed makes a 900 MB file a twenty-minute wait.
Doing it locally, start to finish
Our Auto Caption Generator runs the whole pipeline in the tab: it pulls the audio with ffmpeg.wasm, transcribes with a Whisper model, and shows the captions live over your video. You can fix any misheard word inline, pick a caption style, then export an .srt, a .vtt, or a captioned video.
- ✦The model downloads once on first use, then it is cached — later videos start immediately.
- ✦Longer videos take longer, because your own machine is doing the work rather than a rented GPU.
- ✦Clean audio matters far more than video quality. A decent mic beats a 4K camera for accuracy.
- ✦Always read the transcript before exporting. Names, jargon, and acronyms are where every speech model slips.
Two adjacent tricks worth knowing
If you only need the words — show notes, a blog post, a searchable archive — skip video entirely and use the Audio Transcriber, which handles files or live microphone input. And if you are turning a captioned moment into a shareable loop, the Video to GIF tool picks up where captioning leaves off. Need just the audio for a podcast cut? The MP3 Extractor pulls it out with optional trimming.
Can I add subtitles to a video for free?
Yes. Browser-based tools using Whisper can transcribe and caption video at no cost, with no account and no watermark, because the processing happens on your own device instead of a rented server.
What is the difference between an SRT file and burned-in captions?
An SRT is a separate text file with timestamps that players display over your video — viewers can turn it off and platforms can index it. Burned-in captions are drawn permanently into the video pixels, which is necessary on platforms that do not accept subtitle files.
Is it safe to caption a private or confidential video?
It is if the tool runs locally. When transcription happens in your browser, the file never leaves your device, so there is no server copy and no retention policy to trust. Upload-based services necessarily hold a copy of your footage.
How accurate is automatic captioning?
Modern speech models handle clear speech very well, but they still miss proper nouns, technical terms, and heavy background noise. Treat the output as a strong first draft and read it through before publishing.
Tools in this guide
Try the tools ✦
Free browser tools that never upload your files.