Auto Subtitle Generator – Free AI Captions for Any Video
AIThis subtitle generator listens to your video with Whisper, an open speech model running in your browser, and writes captions timed word by word. Check every word, pick a caption style, then burn the captions into the video, add them as a subtitle track or download an SRT file. Nothing is uploaded and there is no watermark.
- Runs on your device
- Whisper AI
- WebCodecs · GPU
- Works offline
Updated
How to add auto captions to a video
Step 1: Drop a video or an audio file on the box above and set the Spoken language, or keep Detect automatically.
Step 2: Choose a Speech model and press Generate captions. The model downloads once after you confirm.
Step 3: In Review, click any word to correct it; split or merge captions and nudge their timing where needed.
Step 4: In Style, choose Plain, Boxed, Karaoke or Pop, then the font, size, colours and position.
Step 5: In Export, burn the captions in, add a subtitle track, or download SRT, VTT or ASS.
Good to know
Settings explained
- Spoken language
- Picking the language skips detection and improves accuracy on short clips.
- Speech model
- Downloaded once after you confirm, then cached for offline use. Bigger models are more accurate but slower.
- Fastest: from 44 MB · good for clear speech
- Balanced: from 80 MB · recommended
- Accurate: from 252 MB · slower
- Best: from 566 MB · needs WebGPU
- Caption style
- Karaoke highlights each word as it is spoken; Pop shows one punchy word group at a time.
Formats
How it runs
- Whisper speech model
- Speech recognition uses an open Whisper model. It downloads once after you confirm its size, then runs on your device — on the GPU via WebGPU when available.
- Hardware video engine
- Video is decoded and encoded by your device’s media hardware through WebCodecs — usually several times faster than real time.
- GPU canvas
- Pixels are drawn and composited with the GPU-accelerated canvas.
Burn in, subtitle track or SRT file: which output?
Burn captions in when the video goes to Reels, TikTok or Shorts, where viewers cannot switch subtitles on. Add a subtitle track to keep the picture untouched and let viewers turn captions on or off. Download a file when another platform or editor will display them. Already have an SRT? Use add subtitles to video or burn subtitles into video instead.
| Export option | What you get | Picture quality | Best for |
|---|---|---|---|
| Burn in | An MP4 with the captions drawn into every frame | Re-encoded at high quality; the sound is copied when possible | Reels, TikTok, Shorts, any player |
| Subtitle track | Your video plus a WebVTT track viewers switch on with CC | Copied without re-encoding when the container allows | Closed captions, archives, VLC and similar players |
| File only | An SRT, VTT or ASS file | Untouched | YouTube uploads, video editors, styled ASS subtitles |
| Audio file in, Make video | A 1280×720 MP4 with captions over a dark background | New video | Podcast and voice-note clips |
Caption styles and settings explained
Captions are built from the word timings, not from fixed time slices. A new caption starts at the end of a sentence, after a pause longer than 0.7 seconds, or when the line would pass 42 characters, the words-per-caption limit or 6 seconds on screen. Captions in Arabic and Urdu are laid out right to left.
| Setting | Choices | What it does |
|---|---|---|
| Style | Plain, Boxed, Karaoke, Pop | Boxed puts a dark box behind the words; Karaoke highlights each word as it is spoken; Pop shows one punchy word group at a time |
| Words per caption | 1 to 12 (default 5) | Fewer words suit vertical video; more suit wide shots and lectures |
| Font | Sans, heavy display, serif, monospace, rounded | Bundled with the page, so the burned-in result looks like the preview |
| Size and position | Size slider; top, middle or bottom | Middle keeps text clear of the buttons on Reels and TikTok |
| Colours | Text colour and highlight colour | The highlight colours the active word in Karaoke and Pop |
| UPPERCASE | On or off | Common for short-form captions |
How accurate are auto captions?
Clear speech in a common language comes out mostly right, and the errors you will see are names, brand words and numbers. That is why the Review step exists: click a word to retype it (an empty word is deleted), split a caption at any word, merge it with the next one, or nudge it by 0.1 second. The caption under the playhead stays highlighted while the video plays.
A bigger model fixes more than editing does. Balanced downloads 80 MB without WebGPU; Accurate (252 MB) copes better with accents, music and background noise.
What are closed captions?
Closed captions are captions the viewer can turn on or off, stored as a separate track or file. Open captions are burned into the picture and always visible. On this page, Subtitle track and File only make closed captions; Burn in makes open captions.
Strictly, captions also describe sounds, such as [music] or [laughter], for viewers who cannot hear, while subtitles only carry speech. The generator transcribes speech, so add any sound descriptions yourself in Review.
Add captions to Instagram, TikTok and YouTube videos
Each platform treats captions differently, so pick the export that matches where the video is going.
Instagram Reels and TikTok
Burn the captions in. Pop or Boxed with 2–4 words per caption reads well on a phone, and Middle keeps the text clear of the like and comment buttons. Make the video vertical first if it is landscape.
YouTube
Choose File only and upload the SRT in YouTube Studio under Subtitles. Viewers can switch the captions off, and the text stays editable.
Facebook and LinkedIn
Both accept SRT caption files for uploaded videos, which keeps the picture clean. Burn in only if the video will be reposted where files are not supported.
Podcasts and audio
Drop the audio file and choose Make video: you get a 1280×720 MP4 with your captions over a dark background, ready to post as a clip.
Why use this subtitle generator
- Private. The speech model runs on your device, so unreleased videos and client footage are never uploaded.
- No watermark, no credits. Every export is clean and free, however long the video.
- Word-level timing. Captions follow the speech word by word, which is what Karaoke and Pop styles need.
- Your fonts in the file. Burned-in captions use the same bundled fonts as the preview.
- Works offline once the speech model is cached in your browser.
Tips & troubleshooting
Landscape video for TikTok or Reels? Crop it to 9:16 around the speaker with auto reframe video or the horizontal to vertical video converter, then caption the vertical file.
Need the captions in another language? Download the SRT, run it through translate subtitles, then burn the translated file with burn subtitles into video.
For heavy timing work on an existing file, the subtitle editor online has a waveform and per-cue timing.
Short clips: set the Spoken language yourself; automatic detection has little to go on in a few seconds.
Want text instead of captions? Video to text gives a readable transcript with paragraphs.
Auto Captions (AI Subtitle Generator) FAQ
Is there a free subtitle generator with no watermark?
Yes, this one. Captions are generated by a speech model running in your browser, so there are no credits, minute limits or watermarks, and the video is never uploaded. The only download is the model itself, once, after you confirm its size.
How do I add captions to a video for free?
Drop the video here, press Generate captions, check the words in Review, pick a look in Style, and choose Burn in under Export. You get an MP4 with the captions drawn into the picture at the original resolution, with the original sound copied wherever the format allows.
How do I generate an SRT file from a video?
Generate the captions, review them, then choose File only in Export and download SRT. VTT and styled ASS files are available from the same step. The SRT keeps the timing of every caption, so YouTube or a video editor can use it directly.
How do I generate subtitles from audio?
Drop an MP3, M4A or WAV file instead of a video. You can download the captions as SRT, VTT or ASS, or choose Make video to get a 1280×720 MP4 with the captions over a dark background, which works for podcast clips.
How accurate are auto captions?
Mostly right on clear speech in a common language; names, numbers and brand words are where mistakes appear. Every word can be corrected in Review before export, and the Accurate model does better with accents and background noise than the default.
How do I add captions to Instagram videos?
Burn them in: Reels have no place to upload a caption file. Use Pop or Boxed, 2–4 words per caption and the Middle position so the text stays clear of Instagram’s buttons, then post the MP4 from your phone.
Which languages can it caption?
Whisper recognises 99 languages and can detect which one is spoken. The language menu lists the 20 most used, from English, Spanish and Hindi to Arabic, Japanese and Chinese. Accuracy is best for widely spoken languages and clear recordings.