A script timer is not captions
A script timer estimates spoken duration from word count. It does not transcribe, time cues, or burn captions into a video.
A script timer is not captions. It is arithmetic: word count divided by a words-per-minute pace, plus a clip-count table. It does not transcribe a video, it does not place cues, and it does not burn words into pixels. If you searched "script timer" because you needed captions on a Reel, you have the wrong object.
The Versely timer is the script timing estimator. It runs in the browser. Nothing is sent anywhere. It never spends a credit. The sibling page, Script Timing Estimator in the browser is not a generate, is why you should not pay Veo to do a sum. This page is why you should not confuse that sum with a caption burn.
Duration is not a cue list
Paste a script. Pick a pace. Read the spoken duration.
Presets on the free tool:
- Careful / tutorial — 120 wpm
- Conversational — 140
- Typical voiceover — 150 (the usual VO target)
- Commercial — 170
- Fast / hype — 190
The slider runs 80 to 220. 150 is a target, not a prediction of a specific speaker. Pauses, product names, and a breath before a line all stretch real speech. If you already have a recording, time that rather than the script.
The second readout is duration divided by 8, 15, 30, or 60 seconds, rounded up. It does not know where a thought ends. It is a fit check against a format: will this narration cover four 15-second clips, or are you writing a 60-second VO for an 8-second hole. Cut on meaning after you know the script is in the right ballpark.
None of that produces:
- A transcript of an existing video
- Start and end times per line
- Word-level timestamps for karaoke captions
- An
.srtor.vttyou can upload to YouTube - Burned-in captions on an MP4
Caption timing is a different clock. It maps words onto this recording. Forced alignment needs audio. Auto captions run speech-to-text and alignment together. A script timer never hears the file. It has not met your take.
Paying a caption generator to "see how long this reads" is the same waste as paying a video model to do the sum, just a different meter. Run the estimator first. Then record or generate speech. Then caption the locked cut.
If the estimator says 40 seconds of speech, you do not have an 8-second Veo 3.1 problem (4 / 6 / 8 second chips). You have a script problem, or you need several shots. Inworld TTS 2 will read the 40 seconds if you pay for the characters. It will not make them fit in 8.
Captions are a later pass on a locked picture
Order of operations for a talking short:
- Write the line you actually need.
- Paste it in the estimator at 150 wpm, then at the pace you will really speak.
- If it overruns the slot, cut the script — not the model budget.
- Generate or record picture and voice only when the arithmetic says it fits.
- Lock the cut.
- Caption: transcribe (or paste a known script into an alignment workflow), style, burn in for the feed.
Step 2 is the script timer. Step 6 is the AI caption generator. They do not share a button.
Burn-in is pixels. A later trim is a recaption. That is why the timer belongs before you generate, and the captioner belongs after you lock. Reversing them is how a "quick recut" becomes a second caption bill and a timer you never needed because the audio already existed.
Sidecar captions (.srt, .vtt) are still not a timer. They are a track a player draws. Closed captions name that switch. Short-form almost never wants them; a muted Reel will not enable a track. If a client asked for a "script timer" and a YouTube file, they may have meant a caption file. Ask. Then export a transcript from get a video transcript and author the VTT. Do not send them a WPM table.
SDH and forced narrative are content jobs on top of a real caption file. A timer will not invent [music] or pick the one German line that must appear. CC vs subtitles vs forced narrative is that split.
Reading speed for captions is related to WPM and is still not the estimator. Broadcast practice often uses a band around 160–180 wpm for adult same-language captions; a cue that vanishes before you can finish it is a timing error on the video, not a script that is "too long" in the abstract. Watch the burned cut on mute. The estimator cannot watch.
What the free tool will not do, on purpose
It will not transcribe a recording.
It will not know that "Q3" is one token and three syllables.
It will not split on sentences for cues — that is a different free tool if you need a scratch .srt from a script.
It will not caption in 165 languages.
It will not preview Glass versus Classic on your footage. Caption style preview is a 480p pass on a real clip, free with a cooldown, in the editor. Different object, still not a timer.
It will not add a hook overlay. Overlay is copy you wrote. Captions are speech. The timer is neither.
Generation on Versely always costs credits. There is no free generate tier. Arithmetic on your own text is not a generate. That sentence is the product boundary. Caption burns are generates in the sense that they cost credits and produce a new MP4. The estimator is upstream of all of them.
Search "script timer" also catches people building a teleprompter clock. This tool is not a teleprompter. It will not scroll. It will not count in from 3. It will tell you the read is 12 seconds so you stop writing a paragraph for a 8-second hole.
Keep the script next to the take as text. That is ordinary AI asset versioning. When the line changes, re-run the estimator before you re-buy TTS or lipsync. Do not recaption a 40-second burn to discover it never fit.
The test
Do you have a script and a duration question? Estimator. Zero credits.
Do you have a video with speech and a mute-feed problem? Caption generator. Burn in.
Do you have a script and no picture? Timer first, then a picture row, then captions. Skipping to captions will fail because there is nothing to burn onto.
Do you have a client saying "send the script timer"? They might mean the estimator, a teleprompter, or captions. Send the WPM result plus a question: "Do you also need words on the video?"
A script timer is a sum. Captions are a render. Use both. Do not merge the names.
Time the script in the estimator, then caption the locked cut in the caption generator. Sign up when the words already fit the slot and the next click is a burn-in, not another word count.
FAQ
Can the script timer burn captions onto my video?
No. It never sees a video. It counts words and divides by WPM. Burn-in is the caption generator, after the cut is locked.
Is the estimator the same as caption timing?
No. Caption timing maps lines onto a recording. The estimator predicts how long a script would take to speak. One is a clock on a file. The other is a forecast on text.
Should I caption first to learn the duration?
No. Duration from a script is a free division. Captioning is a credit render and it assumes picture. Estimate, cut the script, generate, lock, then caption.