Produce a standard .srt subtitle file, with numbered cues and precise timecodes, straight from the speech in your media. Nothing is uploaded, and there’s no account or watermark.
Load the media you need an .srt for: MP4, MOV, WebM or MKV video, or MP3, M4A, WAV or OGG audio.
Set the language and accuracy, then press “Generate subtitles” to have each phrase recognised and timed.
Under “Line layout”, keep “Max characters per line” at 42 and “Lines per subtitle” at 2, or change them and press “Re-split subtitles” to rebuild the cues.
Edit any text or “Start” and “End” time, then press “Download SRT” to save the numbered file.
SRT, short for SubRip Text, is the most widely supported subtitle format. It is a plain text file made of blocks called cues. Each cue has three parts: a sequence number on its own line, a timing line such as 00:01:02,500 --> 00:01:05,000, and one or two lines of text. A blank line separates one cue from the next.
The timecode is hours, minutes, seconds and milliseconds, written HH:MM:SS,mmm, and the separator before the milliseconds is a comma. That comma is the detail people most often get wrong when writing SRT by hand: with a dot, many players ignore the cue. filetidy writes the numbering and timecodes for you from the speech recognition, so the structure is always valid.
SRT has no official specification and almost no styling. Some players honour basic tags for italics or colour, but most platforms strip them, so filetidy writes clean text only. The SRT is saved as UTF-8 with a byte-order mark, which helps older players and TVs display accents and non-Latin scripts.
Readable subtitles matter as much as correct ones. A common rule is at most 42 characters per line and two lines per cue, shown for at least about a second and no more than about seven. filetidy's automatic splitting follows those limits, cuts at sentence punctuation and pauses, and balances the two lines of a cue.
Windows
Notepad can fix wording, but it cannot create timecodes for you. Use it for corrections after generating the file.
Mac
Do not change the commas, arrows or numbering; players reject or skip blocks whose timecode line is malformed.
VTT (WebVTT) is SRT's web-oriented cousin. A VTT file starts with the line WEBVTT, uses a dot instead of a comma before the milliseconds (00:01:02.500), and can carry positioning, styling and notes. It is the format web browsers use for the track element of an HTML5 video, so it is what you need for a video embedded on your own website.
Everywhere else, SRT is the more compatible choice. Upload platforms, desktop players, televisions and editing software accept it almost universally, and its simplicity means fewer things can go wrong. When a platform lists both, either works.
You don't have to choose before generating: the same subtitles can be saved with “Download SRT” and “Download VTT”, and the timing is identical in both.
A number such as 12, then a line like 00:00:41,200 --> 00:00:44,050, then the subtitle text on one or two lines, then an empty line. The numbers count up from 1 in the order the subtitles appear.
That is simply how SubRip, the program that gave the format its name, wrote its times, and players have expected it ever since, so the milliseconds always follow a comma in SRT and a dot in VTT.
Yes. Under “Line layout”, set “Max characters per line” and choose 1 or 2 “Lines per subtitle”, then press “Re-split subtitles”. The cues are rebuilt with new timings while keeping any text you have corrected.
Yes. An MP3, M4A, WAV or OGG file gives a normal SRT, useful for podcasts published as video, audiograms or adding captions to a video edited elsewhere with the same soundtrack.
The file is UTF-8 with a byte-order mark, the most widely recognised way to store any script. Modern players and platforms read it without settings; very old devices may need their subtitle encoding set to UTF-8.
YouTube Studio, Facebook, LinkedIn and most video hosting services accept SRT for captions. Desktop players like VLC load it from the subtitle menu or automatically when it has the same name as the video.
Last updated October 28, 2026
Choosing the language is faster and more accurate than detection, especially for short clips.
Speech recognition makes mistakes, especially with names, numbers, music and several people talking at once. Dialects such as Moroccan Darija are recognised less reliably than Modern Standard Arabic. Read the subtitles before you publish them.
Need long videos or the best accuracy?
AI subtitles Pro will use Whisper large-v3 on our servers: about an hour of video in a few minutes, better accuracy on accents and dialects, files up to 3 hours. Only compressed audio is sent, never the video.
See Pro and join the waitlist