Sync your video script to the recorded voiceover and get SRT captions
You wrote the script, recorded the voiceover, and now you need captions that appear when you say each line. Upload the audio or video with the script and you get SRT and VTT files that keep your wording and take their timing from your delivery. A recording of up to 10 minutes is free in full; a longer one is $5, paid once, for up to 6 hours.
- Script pasted into the box or uploaded as TXT or DOCX, kept word for word
- Times follow your real pacing, not an even split of the running time
- Captions of up to two lines of about 42 characters, broken at sentence and clause boundaries
- Lines you improvised but never wrote down get no caption
- SRT and VTT for YouTube, Premiere Pro, DaVinci Resolve and other editors
From a written script to captions without retyping
Narrators and video creators usually have the text before the audio exists. The slow part comes later: pasting lines into a caption editor and dragging each one until it matches the take. Here the script itself becomes the caption text. A speech recognition model listens to your take and notes when each word is said, and code then lines up your script against those words. Phrases that occur only once serve as fixed points, and the words between them are paired one at a time.
Because the matching happens word by word, an uneven read still lines up: a long pause before a punchline, a quick product rundown, a slow and calm intro. The captions follow the take you kept, not the take you planned.
When the recording differs from the script
- An ad-lib or aside that is not in the script is simply not captioned, and the next scripted line still starts on time
- A word you swapped or skipped while reading takes its timing from the words around it, but the caption shows the script version, so edit the script if you want captions to match what you said
- A scripted paragraph you cut from the final edit has no speech to attach to and gets crowded next to its neighbors; delete it from the script first
- Music beds, sound effects and silent B-roll are passed over and do not push later captions out of place
Deciding where each caption breaks
Every line break in your script is respected. If you want one caption per sentence, put one sentence per line. A line longer than about 84 characters is broken at a sentence end or a comma and laid out as two rows of about 42 characters, a size that reads comfortably on a phone screen.
Scene headings, camera notes and markers such as "[B-roll: city at night]" are never spoken. Take them out before you upload, or they will turn up as captions squeezed in between your real lines.
Checking the timing before you pay
For a voiceover longer than 10 minutes, the opening part is timed on screen first, so you can see whether captions start where you start speaking. If they do, one payment covers the full recording. You can correct any caption on screen, and the downloaded files use your edit. Voiceovers in Amharic, Tigrinya, Somali, Dari, Kurdish, Burmese or Pashto are timed by a general AI model and land less precisely, so look at the free part closely in those languages.
How the tool works, what it costs and how your files are handled is on the Sync Transcript to Audio page; prices for every tool are on the pricing page.
Last updated
Questions and answers
Can ChatGPT sync my script to my voiceover?
For a short clip where you already know the times, a chat assistant can lay a script out as SRT without trouble. It cannot listen to a long voiceover and tell when each line begins, so the timestamps it writes on its own are estimates that slide away from your voice as the video goes on. This tool reads the times from the recording itself.
Do I need to clean up my script first?
Remove anything you did not say aloud: scene headings, shot notes, paragraphs you cut in the edit. Small differences, such as a dropped "so" or one word said differently, can stay.
Which files can I upload?
The recording can be MP3, M4A, WAV, MP4, MOV or another common audio or video format, or a public Google Drive, Dropbox or direct link. Paste the script or upload it as TXT or DOCX.
Will captions keep my spelling of names and brand terms?
Yes. The caption text is your script, so product names, unusual spellings and punctuation appear exactly as you typed them. Speech recognition only supplies the timing.
How much does it cost?
A voiceover of up to 10 minutes is free, downloads included, with no account. A longer recording costs $5 once and can run up to 6 hours. Unlimited, at $19 a week, raises that to 12 hours per job.
Is my unreleased video private?
Only you can open your job. The recording goes to an AI model provider only to be processed and is not used to train models. Jobs started without an account are deleted after 7 days, and you can delete yours sooner at any time.