Drop in a video and get subtitles automatically. Whisper speech recognition runs in your browser, splits the speech into readable subtitles with word-level timing, and lets you check every line on the video before you download SRT or VTT. Nothing is uploaded.
Last updated:
Preparing…
Video preview with live subtitles
0:00 / 0:00
Subtitles
s
Advertisement
Advertisement
How to use the Auto Subtitle Generator
Add your video (or audio file) by dropping it on the box.
Choose the model, language and line settings: 42 characters and 2 lines is the standard; 1 line suits Reels, Shorts and TikTok.
Press Generate subtitles. The first run downloads the Whisper model once, then subtitles appear as each part is recognized.
Play the video and fix any mistakes: edit text and times, split or merge subtitles, or shift them all.
Download SRT or VTT and upload it with your video, or burn it in with Add Subtitles to Video.
How the subtitles are made
Speech recognition. The soundtrack is decoded in your browser, resampled to 16 kHz mono and passed to Whisper in parts of up to 30 seconds, cut at quiet moments so words are not split.
Word timing. With word-level timing on, Whisper aligns every word to the audio using its cross-attention, so each word gets its own start and end time.
Splitting into subtitles. Words are added to a subtitle until the next word would exceed the line limit, the subtitle would last longer than the maximum duration, the speaker pauses for more than a second, or a sentence ends. Commas break long subtitles at natural points.
Line layout and reading time. Two-line subtitles are split near the middle so the lines are balanced, and each subtitle stays on screen at least about a second (without overlapping the next one) so it can be read.
Subtitle settings for each platform
Where
Characters per line
Lines
Max duration
File
YouTube, Vimeo, films and courses
42
2
7 s
SRT or VTT
TikTok, Reels, Shorts (vertical)
20-30
1
3 s
Burned in
LinkedIn, X, Facebook feed videos
32-42
1-2
5 s
SRT (or burned in)
HTML5 video on your own website
42
2
7 s
VTT
Most viewers watch social video with the sound off, which is why short, one-line, bold captions are popular there. For long-form video, standard two-line subtitles are easier to follow.
Editing tips
Times accept 00:01:23,450, 01:23.45 or plain seconds like 83.45.
Split (✂) cuts at the text cursor and divides the time in proportion to the text. Merge (⤓) joins a subtitle with the next one.
A yellow edge marks a subtitle with a long line, an overlap or a reading speed above 20 characters per second. Shorten the text or give it more time.
Out of sync? If every subtitle is early or late by the same amount, use Shift all by. For drifting subtitles, recalculate times with the SRT Time Calculator.
Need just the text? The Audio to Text tool gives you a transcript with paragraphs and a DOCX download.
Frequently asked questions
How do I generate subtitles for a video automatically?
Add the video to this page and press Generate subtitles. The Whisper speech recognition model listens to the soundtrack in your browser and writes timed subtitles, which you can review on the video and download as an SRT or VTT file.
Is it really free and private?
Yes. There is no account, watermark or time limit, and the video is never uploaded: decoding, speech recognition and editing all run on your device. Only the Whisper model (about 40 or 80 MB) is downloaded once and cached by your browser.
What is the difference between SRT and VTT?
Both are plain-text subtitle files with numbered cues and start and end times. SRT uses commas in times (00:00:01,500) and is supported by almost every player and editor. WebVTT uses dots (00:00:01.500), starts with a WEBVTT header and is the format used by HTML5 video on the web.
How many characters should a subtitle line have?
A common standard, used for example in Netflix's English style guide, is at most 42 characters per line, at most two lines, a reading speed up to about 20 characters per second, and each subtitle on screen for between 5/6 of a second and 7 seconds. These are the defaults here.
What does word-level timing do?
With word-level timing Whisper reports when each word is spoken, so subtitles can be cut exactly at pauses and sentence ends and disappear when the speaker stops. Without it, the tool only knows when each sentence starts and ends and estimates the words in between.
Can it add the subtitles into the video?
This page previews how burned-in subtitles look on your video and exports SRT or VTT files. To permanently burn them into the picture, load the video and your SRT into the Add Subtitles to Video tool, or upload the SRT to YouTube, Vimeo or your editor as closed captions.
How accurate are automatic subtitles?
Clear speech in common languages is transcribed well, but names, jargon, music, noise and overlapping voices cause mistakes, especially with the Tiny model. Always watch the video through once and correct the text before publishing.
Are all the tools on FreeToolCabin tested for quality and accuracy?
Yes! Every tool on our site undergoes automated Q/A testing in real browser environments to ensure high precision, correct output, and zero errors. If you happen to spot a bug or have a suggestion, please contact us at [email protected] and our team will review it right away.