Translate to any language
Translate the SRT to 99 languages — one click, keep the timing
Last reviewed October 2026
SRT generator — audio or video in, a timed .srt out
Upload audio or video and get a .srt with speaker labels, timed to the speech, in 99 languages. Or paste a YouTube link.
- 5 minutes free, no account
- 30 more free on signup, no card
- Files up to 5 GB
- Files not used for training
What is an SRT generator?
An SRT generator turns spoken audio into a .srt subtitle file — numbered, timestamped text cues that YouTube, Vimeo, Premiere Pro and most video editors accept. VexaScribe builds the file from audio, video or a YouTube link in 99 languages, placing each cue boundary on word-level timings.
Drop your file here
or click to browse
Audio: MP3 · WAV · M4A · FLAC · OGG · AAC · AIFF · WMA · AMR · OPUS
Video: MP4 · MOV · AVI · MKV · WebM · FLV · WMV
Up to 200 MB · 5 min free preview
What you get
A standard SubRip file: numbered cues, HH:MM:SS,mmm timecodes with the comma the format requires, a blank line between cues, and UTF-8 without a BOM, so accented characters survive and players read the first cue. Uploaded recordings with several speakers carry a label on each cue. The same run also exports VTT, TXT, DOCX and JSON.
1
00:00:04,120 --> 00:00:07,940
Speaker 1: So the first thing we need to cover
is the quarterly numbers.
2
00:00:08,010 --> 00:00:11,600
Speaker 2: Right, and that ties directly
into the hiring question.
3
00:00:11,720 --> 00:00:13,880
Speaker 1: Exactly. Let me pull those up.Speaker labels come from uploaded files. A pasted YouTube link imports the video's own captions, which have no speaker information, so the file comes back without labels. For a multi-speaker video, download it and upload the file instead.
How it works
Nothing to install. Works in any browser.
- 1
Upload or paste a link
Drop in an MP3, WAV, M4A, MP4, MOV or any of 19 supported formats, or paste a YouTube link.
- 2
It transcribes
The language is detected and every word timed. An hour of audio typically takes about 3 minutes; allow longer when it's busy. Short clips still take a minute or two to start.
- 3
Review in the editor
Fix a misheard name, nudge a start time, split or merge a cue.
- 4
Download
Export SRT, VTT or TXT for YouTube Studio, Premiere Pro, Final Cut, DaVinci Resolve, CapCut or VLC.
How the cues are split
A usable SRT depends on cue timing as much as on the words. A 30-second monologue can't be one subtitle, and a two-word reaction shouldn't flash past. Cue boundaries are placed on the word timings the model returns, so a cue starts and ends where the speech does instead of drifting further out with every cue.
| Rule | Value |
|---|---|
| Characters per cue | Up to 80 |
| Target cue length | Up to 7 seconds |
| Extended for readability | Up to 10 seconds |
| Worst case, long unbroken speech | 14 seconds |
| Reading speed target | 15 characters per second |
| Preferred break point | A pause of 0.4 seconds or more |
| Gaps between cues | Under 83 ms closed, so text doesn't flicker |
For reference, the Netflix Timed Text Style Guide caps adult subtitles at 17 characters per second and 7 seconds per cue. The targets above sit inside both; cues stretched toward 10 or 14 seconds are the ones to split in the editor.
Before you publish
No automatic transcript is reliably perfect, and the slips are predictable: names and brand names, homophones, overlapping speakers, and lyrics under music. How often depends on the audio; our Whisper accuracy benchmark measures it from clean audiobooks to phone calls. Do one pass before you publish. The SRT editor opens the file in a browser tab with reading speed and line length flagged per cue; click a cue to fix the text or timing, and re-export.
Frequently Asked Questions
How do I generate an SRT file from audio or video?
Upload the file or paste a YouTube, TikTok or Instagram link. Uploads accept 19 audio and video formats, including MP3, WAV, M4A, MP4, MOV and MKV, up to 5 GB with no length limit. An hour of audio typically takes about 3 minutes on Standard and 1 on Premium. Review the cues in the editor, then download the .srt.
Is it free?
Try 5 minutes right here with no account. Sign up — no card — for 30 free minutes and files up to 5 GB. Paid plans start at $2/month for 200 minutes, and every plan includes unlimited SRT export with no per-export fees.
Why don't my subtitles have speaker labels?
Speaker labels come from uploaded files. A pasted YouTube, TikTok or Instagram link imports the platform's own captions, which carry no speaker information. For a video with several speakers, download it and upload the file instead.
How long can each subtitle cue be?
Up to 80 characters, with a target of 7 seconds at about 15 characters per second. Cues prefer to break on a pause of 0.4 seconds or more. A cue can stretch to 10 seconds for readability, and long unbroken speech can produce one as long as 14 seconds; split those in the editor before you publish.
Does it work in languages other than English?
Yes. 99 languages, detected automatically, and the SRT comes back in the spoken language. You can translate it afterwards with the cue numbers and timecodes kept. For Japanese, Chinese and Korean, cue splitting is less reliable, so check line lengths in the editor before publishing.
Can I edit the SRT file before downloading?
Yes. The built-in editor lets you correct text, adjust start and end times, and split or merge cues, with the audio alongside. Changes save as you go — download the .srt when you're done.
Will the SRT import into Premiere, Final Cut, or DaVinci Resolve cleanly?
Yes. It is standard SubRip — numbered cues, HH:MM:SS,mmm timecodes with a comma, UTF-8 without a BOM. Drag it into the timeline in Premiere Pro, Final Cut Pro or DaVinci Resolve, or upload it to YouTube Studio, CapCut or VLC.
I already run Whisper locally. What does this add?
Standard transcription runs OpenAI Whisper Large-v3, so the text will be familiar. What it adds: cues split on word timings rather than fixed-length chunks, speaker labels without a separate diarization step, no GPU, an editor for fixes, and SRT, VTT and TXT from one run. To keep running it yourself, see our Whisper install guide.
Why isn't it unlimited and free?
Because transcription costs real money to run, and someone pays for it either way. Tools that are unlimited and free usually cover it by capping quality, showing ads, or keeping rights to your audio. We would rather be straightforward: 5 real minutes with no account, 30 minutes free when you sign up, and plans from $2/month after that. Your files are not used to train models, and we do not keep rights to your audio.
Generate your first SRT file
Try 5 minutes right here, no account. Sign up — no card — for 30 free minutes and files up to 5 GB. Plans start at $2 a month.
What else you can do with your transcript
AI Summary
Get the 5 key moments — skip the full read
AI Chat
Ask questions like "when did they mention pricing?"
Export anywhere
DOCX · VTT · JSON · CSV — one processing