🎬 Subtitle Generator

Upload a video and download a ready-to-use SRT or VTT subtitle file. Timings and line breaks are done for you.

🎬

Drag & drop a video or audio file or click to browse

MP4, MOV, MKV, WEBM, AVI, MP3, WAV, M4A — up to 500 MB

Generate SRT subtitles from any video, free

Drop in an MP4 and get back a subtitle file you can upload straight to YouTube, load into Premiere or DaVinci Resolve, or drop next to a video file so VLC picks it up automatically. The audio is transcribed by a speech recognition model, cut into cues, timed against the original, and wrapped to the line lengths subtitle players expect. You get three files from one upload: SRT, WebVTT and a plain-text transcript.

Which format do you need?

FormatUse it forNotes
.srtYouTube, Vimeo, Facebook, VLC, Premiere, ResolveThe safe default. Almost everything reads SRT.
.vttWeb video — the HTML <track> elementRequired for subtitles in a browser video player.
.txtShow notes, blog posts, search, translationNo timecodes — just what was said.

How it works

  1. Upload. Any container ffmpeg can read: MP4, MOV, MKV, WEBM, AVI, plus audio files like MP3, WAV and M4A.
  2. The audio is separated from the video. Only the sound is needed, so the video track is discarded immediately and never leaves the queue.
  3. Speech recognition runs. The language is detected automatically unless you pick one.
  4. Cues are built. Long sentences are split across several cues, lines are wrapped to 42 characters, overlaps are corrected and empty cues dropped.
  5. Download. SRT, VTT and TXT, from the same transcription.

What makes a subtitle file good

Most automatic subtitle tools give you accurate words in an unreadable shape: one enormous cue per sentence, lines that run off the side of the screen, and cues that overlap the next one. Those files technically validate and are miserable to watch. This generator applies the conventions broadcast subtitlers use:

Limits

Frequently asked questions

What is an SRT file?

SubRip Subtitle — a plain text file listing numbered cues, each with a start and end timecode and the words spoken in that window. It is the most widely supported subtitle format there is: YouTube, Vimeo, Facebook, VLC, Premiere Pro, Final Cut and DaVinci Resolve all read it. You can open one in any text editor and fix a typo by hand.

How accurate are the subtitles?

Accurate enough that editing is faster than typing from scratch, which is the honest benchmark for any automatic transcription. Clear speech with one speaker and little background noise transcribes very well. Accuracy drops with heavy accents, several people talking over each other, music under the dialogue, or specialist vocabulary and proper nouns. Always read through the result before publishing — names and technical terms are where mistakes cluster.

Do I need to pick the spoken language?

No. The language is detected from the audio. Choosing it explicitly helps in two situations: very short clips, where there is little audio to judge from, and recordings that mix two languages, where detection may settle on the wrong one. If your result comes back in the wrong language, pick the right one and run it again.

Which video formats can I upload?

Anything ffmpeg can open, which covers essentially every format in normal use: MP4, MOV, MKV, WEBM, AVI, FLV, WMV and more. Audio files work too — MP3, WAV, M4A, AAC, FLAC, OGG and Opus. The video track is irrelevant to the result, so a phone recording and a ProRes master transcribe the same.

How long does it take?

Roughly one minute of processing for every seven minutes of video, plus the upload. A ten-minute video is usually done in under two minutes. Jobs run one at a time, so if someone else is ahead of you the page will tell you your position in the queue rather than leaving you guessing.

Can I close the tab while it runs?

Better not to. The job keeps running on the server, but the link to your result lives only in that tab — there is no account to come back to. If you reload or close it, you will need to upload again.

Is there a limit on video length?

Sixty minutes per file, and 500 MB. Free transcriptions are capped at twenty minutes per file on top of your daily allowance. For anything longer, split the video first — most editors can export a section, and the timecodes in each SRT will start from zero for the section you exported.

What happens to my video?

The audio is extracted, the original upload is deleted straight away, and the extracted audio is deleted as soon as the transcription ends. The text result is kept for a few hours so you can return to the tab, then removed automatically. Nothing is published, indexed or shared, and there is no account holding your files.

Can I edit the subtitles afterwards?

Yes, and you usually should. SRT is plain text, so any text editor will do for fixing a word. For adjusting timing, use a subtitle editor. If the whole file is out of sync by a constant amount — a common problem when the video was re-encoded at a different frame rate — a shift tool will fix it in one step rather than cue by cue.

Why are the lines split the way they are?

Subtitle lines are held to 42 characters and two lines per cue, which is the convention broadcast subtitlers use and what most players are laid out for. Where a sentence does not fit, it becomes several cues rather than one long one, with the time divided between them. The result reads at a natural pace instead of flashing a paragraph on screen.

Can I use these subtitles on YouTube?

Yes. In YouTube Studio open your video, go to Subtitles, choose the language and select Upload file, then pick the SRT with "With timing". YouTube reads the timecodes as they are. This is worth doing even though YouTube generates its own captions: an uploaded file is treated as authoritative, is used for search indexing, and does not carry the errors automatic captions introduce.

What is the difference between SRT and VTT?

They hold the same thing in slightly different syntax. WebVTT starts with a WEBVTT header, separates seconds from milliseconds with a full stop rather than a comma, and does not number its cues. VTT is what the HTML <track> element requires, so it is the one you need for video on a web page; SRT is what nearly everything else expects. Both are produced here from the same transcription, so you never have to convert between them.

Does it work for languages other than English?

Yes — more than ninety, including Portuguese, Spanish, French, German, Italian, Dutch, Russian, Arabic, Hindi, Japanese, Korean and Chinese. Accuracy varies by language, since the model saw far more English than anything else, but the major European languages transcribe well. The subtitles come back in the language that was spoken; this tool transcribes, it does not translate.

Can it translate the subtitles into another language?

Not here — this tool writes down what was said, in the language it was said in. To translate a finished subtitle file, use the SRT Translator on kliptools.com, which takes the SRT you download here and returns it in another language with the timings untouched.

Does it separate different speakers?

No. There is no speaker labelling, so an interview comes back as one continuous transcript rather than tagged by who was talking. For subtitles this rarely matters, since viewers can see who is speaking. If you need speaker labels, add them by hand in a subtitle editor.

Do I need an account?

No. There is nothing to sign up for and no email required. Each visitor gets a free allowance of transcription minutes per day; past that a KlipTools credits code pays for the rest, and the code is entered on this page rather than tied to an account.