How to add captions to videos free: AI, auto-sync, burn-in

Why captions matter more than you think

Captions aren't just for accessibility (though they're critical for deaf viewers). They boost engagement — videos with captions get 80% more views on YouTube. They're searchable, which means YouTube's algorithm indexes your spoken words. And they let viewers in noisy environments, at work, or on mute still follow your content. On social media, captions are the difference between a video getting scrolled past and someone stopping to watch.

The barrier used to be cost and effort: hiring a transcriber ($0.50-$2 per minute), or manually captioning yourself frame by frame. Now AI transcription is free and available in-browser, making captions accessible to everyone.

Soft captions vs. hard captions: which should you use?

Soft captions (SRT files or embedded as tracks) are separate from the video. Viewers can toggle them on or off. YouTube, Vimeo, and most streaming platforms support them natively. The downside: not all platforms display them, and some viewers don't know the toggle exists.

Hard captions (burned-in) are permanently baked into the video file. They're always visible, on every platform, on every device. The downside: you can't edit them without re-encoding the video, they take up bitrate space, and they can't be toggled off (bad for viewers who don't need them).

For accessibility and SEO: use soft captions. For sharing on social media platforms that don't support caption tracks: use hard captions. The ideal workflow is to create soft captions, then burn them in as a separate export for social media distribution.

How AI transcription actually works (and where it fails)

Modern AI transcription (like Whisper from OpenAI) is 95%+ accurate on clear audio. It handles most accents, background noise, and technical jargon better than older speech-to-text systems. But it still struggles with:

• Background noise: Loud music, crowd chatter, or traffic can confuse the model into transcribing noise as words.
• Accents: Non-native English speakers are transcribed less accurately than native speakers. The accuracy gap is shrinking but still exists.
• Proper nouns: Brand names, person names, and place names are often transcribed phonetically instead of correctly.
• Technical jargon: Specialized vocabulary (medical, legal, software terminology) gets transcribed as homophones (e.g., "Python" as "python").
• Filler words: Accurate models actually transcribe "um," "uh," and stutters — which you then have to manually clean up.

The practical takeaway: let AI do the heavy lifting, then spend 5 minutes reviewing for obvious errors. Don't trust it to be 100% correct without a human pass.

The free workflow: generate, sync, embed

Step 1: Transcribe — Upload your video to a free AI transcription tool (WavyVid, or browser-based Whisper). It generates a transcript or subtitle file (usually SRT format). This takes 1-2 minutes for a 5-minute video in-browser.

Step 2: Review and edit — Open the generated SRT file in any text editor. It's plain text with timestamps. Fix typos, proper nouns, and obvious transcription errors. This takes 5-10 minutes depending on video quality and length.

Step 3: Sync (optional) — If the timestamps are off, you can manually adjust them in the SRT file. Most tools auto-sync based on the audio, so manual adjustment is rare.

Step 4: Embed soft captions — Upload your video and the SRT file to YouTube, Vimeo, or another platform. They handle the rest. Or use a free tool to burn captions into the video as a secondary export.

Total time: 10-20 minutes for a 5-minute video. That's faster than manual captioning and the results are 95% accurate.

Where to add captions for free

YouTube: Upload your video, go to Subtitles, and auto-generate. YouTube's speech-to-text is good but not perfect. You can edit the generated captions or upload an SRT file. Free, built-in, no uploads to third-party sites.

Browser-based AI: Tools like WavyVid or OpenAI's Whisper (run in-browser) transcribe without uploading. Your video never leaves your device. Takes 1-2 minutes per 5-minute video.

Paid-but-free-tier: Descript and CapCut both offer free tier transcription with a monthly-minute cap. Descript is 300 minutes/month free; CapCut is more generous but comes with watermarks. Both are faster and more accurate than YouTube's auto-captions.

Command-line (nerdy): Run FFmpeg with Whisper model locally. Completely free, but requires installation and technical knowledge.

Does adding captions actually increase views?

Yes, but not evenly. On YouTube, videos with captions rank higher in search and recommendations because the algorithm can index the spoken words. On TikTok and Instagram Reels, captions are essential because viewers often watch on mute. On Podcasts embedded on websites, captions help with SEO and accessibility but don't directly boost platform metrics.

The real win is in retention: viewers without audio (at work, in bed, in a crowded room) can still follow your video, so they watch longer instead of scrolling past.

The pragmatic caption strategy

1. Let AI transcribe your video in-browser (free, private, fast).
2. Spend 5 minutes reviewing for errors.
3. Upload soft captions to YouTube or your platform.
4. Export a version with burned-in captions for social media.
5. Done — and your video is now accessible and discoverable.

You don't need Descript. You don't need to hire a transcriber. AI gets you 95% there, and that's good enough for most use cases.