Auto Subtitle Generator for Video

What the automatic part does

Automatic covers more than transcription. Between choosing the file and seeing lines on the preview, four jobs happen without any input from you, and the progress strip names each one while it runs.

Reading the audio

Your browser decodes the soundtrack and compresses it. The picture is never uploaded.

Sending the audio

Only that compressed audio travels to mydubly's server, over HTTPS.

Detecting the language and writing subtitles

Whisper identifies the spoken language and transcribes the speech with a time for every word.

Timing the lines

The words are grouped into short lines, each starting when the first word of its phrase is spoken.

If you asked for a translation, a Translating stage appears before the last one. What stays with you is judgment: whether a name is spelled right, whether a line reads naturally, and how the text should look on your footage.

Running the automatic subtitles

1. Open the editor and make sure "Write the subtitles automatically with AI" is ticked before you choose a file.
2. Leave Subtitle language on "Same as spoken (auto-detect)", or pick a language under "Translate into".
3. Choose the video. If you aren't signed in, a Google sign-in opens and the job starts the moment you finish.
4. Keep the tab open while the stages run. Stop cancels the run if you picked the wrong file.
5. When it's done, a note tells you how many lines were written and which spoken language was found. Play from the start and read along.
6. Correct, style and render the video, or export SRT or VTT.

Why the lines come out short

Plenty of generators turn each sentence into one subtitle, which on a phone means two or three crowded lines that sit there for several seconds. These lines are built differently from the word timings: about 32 characters each (about 12 for Chinese and Japanese, about 14 for Korean), and none stays up much longer than 2.6 seconds. The text changes at the pace of the voice, closer to live television captions than to film subtitles. And because each line begins on its first spoken word, nothing appears before the speaker opens their mouth.

Short lines suit sound-off scrolling and vertical video. They suit dense lectures less, where viewers like a whole sentence to read. If that's your material, Merge joins neighbouring lines, and the editor flags any result that's too fast to read or longer than 42 characters. Splitting and merging subtitles covers that tidy-up in detail.

Translated subtitles are the exception to the word-timed rule. A translation doesn't line up word for word with the speech, so translated text is divided by length across the phrase, and each line's timing is approximate.

Language detection: when to look twice

Detection listens to the speech and decides the language, so a Hindi vlog, a French interview and an English tutorial all use the same setting. On clear speech in one language it's dependable. Give the result a second look when:

• the video opens with a long music intro, a jingle or silence, leaving little speech to judge from;
• speakers switch between languages mid-sentence, as in many Hindi and English conversations, which produces more errors than a single language;
• the clip is only a few seconds long.

The note after each run names the language that was found. If it's wrong, the lines will be too, and there's no setting in the editor to force a spoken language. For clips with a long instrumental opening, trimming that opening in another app before generating gives detection real speech to work from. Hindi always comes out in Devanagari script; romanised Hinglish isn't produced automatically. Background on how this works: spoken language identification.

Running it again

The AI subtitles panel stays available after the first run. Its button changes to "Replace with AI subtitles" and shows the credit cost for this video. The usual reason to use it is a second language: you published English subtitles and now want a Hindi or Japanese version of the same video. Change Subtitle language and run again.

When the new lines arrive, you're asked whether they should replace the ones you have. The question comes after the run, so the credits are spent whichever answer you give. Edits you made to the old lines don't carry over, so export them as SRT first if you want to keep that version. An account runs one AI job at a time; wait for one video to finish before starting the next.

What it costs

Rate

1 credit per minute of video, with a 5-credit minimum per run

In money

1,000 credits = $1, so a 20-minute video is 20 credits (2¢)

Free credits

100 on your first Google sign-in, then 10 a day for registered accounts, accumulating through the month and resetting on the 1st

Always free

Editing, styling, rendering the MP4 and exporting SRT or VTT

Videos can be up to 2 hours long for the AI step. The uploaded audio and the text are deleted from the server within 30 minutes of delivery.

Automatic isn't the same as finished

A generated draft saves the slow part, typing and timing, but it still needs a reader. Before you publish, check every name and brand, any figure read aloud, and the first and last minute, where music and greetings trip the recogniser most. How the model reaches its guesses, and a quick review routine, are on the AI caption generator page. If the subtitle file is all you need, without preview or burning in, the subtitle file generator is the shorter path; for a file in another language there's translated subtitles.

About the subtitle editor on this page

Opens

MP4, MOV, M4V, WebM, MKV videos your browser can decode

Subtitles from

AI (from the speech, optionally translated into one of 21 languages), an SRT or VTT file, or typing them in

You download

An MP4 with the subtitles drawn into the picture at the original resolution and frame rate, plus SRT and VTT files

Cost

Editing, styling, rendering and SRT/VTT export are free, with no account and no watermark. AI subtitles cost 1 credit per minute (5-credit minimum per video); signing in with Google gives 100 free credits and accounts get 10 free credits a day

Privacy

The video never leaves your device and is rendered in your browser. Only for AI subtitles is the audio sent, over HTTPS, and it is deleted within 30 minutes of delivery

Rendering needs

Chrome or Edge 94+, Safari 16.4+, or Firefox 130+ on a computer. Editing and SRT/VTT export work in any current browser