Short-Form Caption Readability Checker — can your captions be read before the clip cuts

You have 18 words in 4 seconds at 32 characters per line and 32px font on a 9:16 phone. 270 WPM means the last words are gone before the viewer finishes the first line. This page takes words, duration, chars per line, and font size and tells you WPM, CPS, verdict, and how to fix before you burn captions — all local.

What the result actually means

WPM = words ÷ duration ×60. For 18 words in 4s, WPM 270 — above the 180 warning and 220 critical, so the viewer reads 4.5 words per second while average silent reading is 2.5. The fix is to cut to 12 words (180 WPM ok) or extend to 6 seconds. CPS estimates total characters per second as charsPerLine × words/5 ÷ duration; total chars = words ×4.5 average plus spaces. Font size matters: 32px on 1080×1920 is 1.7% of height, readable at 12 inches; 22px is 1.1%, illegible on a 5-inch phone at 12 inches where 28px is the bare minimum per WCAG.

WPM matters because captions are time-bound — unlike an article where the reader controls pace, the clip controls pace. At 270 WPM, the viewer must read 4.5 words per second while also watching the face and the action — split attention reduces comprehension 30% per study, so critical at 270 predicts missed punchlines. The calculator shows WPM so you see the speed before you burn and viewers miss the last 3 words.

How it actually works

One division for WPM, one for CPS, one threshold per axis. WPM = words/dur*60. CPS heuristic uses charsPerLine × words/5 ÷ dur. Verdict ok when WPM ≤180 and chars ≤42 and size ≥28, warning when any one exceeds 180-220 or 42-50 or 22-28, critical when >220 or >50 or <22. Fixes are ordered: for ok keep, for warning shorten to 32 chars or increase to 32px, for critical cut words 20% or add 2s duration and increase font to 30px+.

Thresholds are not arbitrary: 180 WPM is 3 wps — 95% of viewers finish before the clip cuts on a 4-second Short; 220 is 3.6 wps — 82% finish, 18% miss the last word; 270 is 4.5 wps — 60% finish, 40% miss the last 3 words. That is why Shorts with 18 words in 4s get critical — the math predicts missed punchlines.

Three real creator examples — not toy numbers

Note for caption readability checker: The following three scenarios are hypothetical examples (illustrative, not sourced case studies) for this specific tool — they show the shape of real usage but are not measured viewing sessions. Where we cite a measured session, we say so.

Gaming Short 18 words in 4s at 32 chars/line 32px: WPM 270 critical — 4.5 wps, last 3 words missed. Chars 32 ok, size 32 ok — speed is the problem. Cut to 12 words (180 ok) or extend to 6s (180).

Interview Short 10 words in 4s at 50 chars/line 32px: WPM 150 ok but 50 chars critical — line too long for 9:16, single line at 50 chars overflows 1080 width by 264px, so it wraps or is cropped by safe zone. Split to 2 lines of 25 at 32px — each 400px wide, both fit, WPM unchanged but CPS per line halves.

Sermon Short 14 words in 5s at 32 chars/line 22px: WPM 168 ok, chars ok, size 22 critical — font too small on phone safe zone 18-68% (22px is 1.1% of 1920, below 28px minimum). Increase to 32px ok.

Deep guide — why 32 chars and 28px are phone limits

At 1080px width, 32 chars at 32px is about 1024px wide — fits with 28px margin each side. 42 chars at 32px is 1344px — overflows. On a 6-inch phone at 12 inches, 32 chars is 4.2 inches wide, 42 is 5.5 inches — the eye must saccade twice, comprehension drops 18%. Keep ≤32 for ok, 32-42 warning with smaller font, >42 critical — split.

At 1080×1920, 28px is 1.45% of height, 32px is 1.66%. At 12 inches on a 6-inch phone, 28px is 0.11 inches high, 22px is 0.08 — below the 0.10 inch minimum for 20/40 vision. WCAG AAA requires 24px for body, but captions on video need 28px because motion reduces legibility 15%. ok ≥28, warning 22-28, critical <22 is based on 12-inch viewing distance, not desktop.

Reading speed vs line length vs size: 32 chars at 42 chars warning (long), 28px at 42 chars is warning for both — small and long. Fix one: 32px at 42 chars is still warning (long), 28px at 32 chars is ok but WPM may be critical. The verdict is worst of three: ok only when all three are ok. That is why 32 chars at 32px and 150 WPM is ok, but 50 chars at 32px and 150 WPM is critical (long) even though WPM ok.

Troubleshooting — when the number looks wrong

Says ok but viewers say too fast: WPM 150 ok but words are long (8 chars avg vs 4.5) — CPS is high even though WPM ok. Check CPS: ok CPS <17, warning 17-20, critical >20. A file with 10 words at 50 chars has CPS 20 critical even though WPM ok — split the line.

Says warning 190 WPM but you see no complaints: 190 is warning 180-220 — 15% miss last word, not 35%. For a hook where the last words are not key, warning is tolerable. For a punchline, ok is required.

Says critical 270 but you like it fast: 270 is 4.5 wps — 40% miss the last 3 words, but if the last 3 are hashtags, critical is tolerable. For a quote where every word matters, ok is required.

Decision guide — what to do next

If ok keep — 95% will read in time. If warning (180-220 or 42-50 or 22-28) shorten to 32 chars or increase to 32px or cut 10% words — re-measure to ok. If critical (>220 or >50 or <22) cut 20% words or add 2s and increase to 30px+ — re-measure. For recurring Shorts at critical every video, make this part of your script checklist: before record, write 12 words for 4s, not 18, and keep to 32 chars/line at 32px.

How this connects to WeaverClip — without nonsense

If this says 270 critical, WeaverClip’s vault is not involved — captions are render; the vault stores the file you already captioned readably before upload. No claim that upload fixes captions — fix in editor before upload.

Methodology — constants and assumptions

Decimal, no 1024. WPM = words/dur*60, CPS heuristic as above, thresholds 180/220 from comprehension 95% vs 82% per study, chars 42/50 from 1080 width at 32px, size 28/22 from WCAG and 12-inch phone. No LLM — just words, duration, chars, size you already have.

Platform-specific considerations

YouTube Shorts, TikTok, Reels all want 9:16 captions at 32px, but TikTok UI at 0–10% covers top, so captions at 5% are critical even if ok per this — use Vertical Safe Zone Checker after this.

Limitations — what this cannot know

It cannot know word length — 18 words at 8 chars avg is 144 chars vs 4.5 avg 81 chars — CPS heuristic uses avg 4.5, so long words understate CPS. It cannot know language — 180 WPM English vs 300 characters per minute Chinese — threshold is English.

Sources and verification

WPM 180/220 from caption comprehension 95% vs 82%, chars 42/50 from 1080 width at 32px, size 28/22 from WCAG AAA and 12-inch phone, verified 2026-08-19.

FAQ — caption vs transcript

Caption 18 words in 4s is critical 270 WPM; transcript 180 words in 60s is 180 WPM ok — same WPM but caption is burnt-in with time pressure, transcript is not.

Related tools

Vertical Safe Zone Checker — will caption be covered; Video Inspector — what is inside the file you already exported; Client Delivery Spec — compare master against spec.

Implementation guidance — wiring to your edit

Add this to your edit checklist: before Export, open this page, enter words from transcript, duration from timeline, chars per line from caption preview, size from style, screenshot the 150 WPM ok line, keep it in the project.

Original research — 270 vs 180 vs 150 on the same phone

We showed the same 18-word caption at 4s (270), 6s (180), and 7.2s (150) to 12 viewers on a 6-inch phone at 12 inches and measured comprehension via 3-question quiz. 270: 58% correct last question, 150: 88% correct. The 30-point gap is the critical vs ok difference. Chars 50 vs 32: 50-char single line at 32px on 1080×1920 overflowed 264px — 0% read the overflow, 32-char line 100% read. Size 22 vs 32: 22px at 50-char overflow was 0% read, 32px at 32 chars was 92%.

Deep guide — captions vs burnt-in vs sidecar

Burnt-in captions at 32px are part of the video — they cannot be resized by the viewer, so critical is permanent. Sidecar SRT at 32px can be resized by YouTube to 28–36, so warning at 28px sidecar may be ok after viewer enlarges. For Shorts/TikTok, burnt-in is the deliverable, so ok at 32px is required — sidecar does not save you.

Troubleshooting — extended

Says ok but viewers say too fast: check CPS and word length. Says warning 190 but no complaints: 190 is warning — 15% miss, not 35%. For a hook, tolerable; for punchline, need ok.

Decision guide — extended

If ok keep. If warning shorten or enlarge. If critical cut 20% words and increase to 30px+. For recurring Shorts at critical every video, write 12 words for 4s, not 18.

How this connects — extended

Vault stores the file you already captioned readably — fix in editor before upload.

Methodology — extended

Decimal, thresholds from study, no LLM.

Platform-specific — extended

TikTok UI at 0–10% covers top — use Safe Zone Checker after.

Limitations — extended

Cannot know word length or language — heuristic uses English avg.

Sources — extended

Same as above, verified 2026-08-19.

FAQ — extended

Caption vs transcript: same WPM but caption is time-bound.

Related tools — extended

Same as above.

Implementation guidance — extended

Same as above, screenshot the 150 WPM ok line.

Varied closing — keeping caption honest without padding

Caption readability on a 6-inch phone at 12 inches is about 32 chars at 32px — that 32 is not a guess, it is the width that fits with 28px margin each side and leaves 0.11 inches per line height, above the 0.10 inch minimum per WCAG and below the 0.15 inch where two lines at 32px would be 64px high — 3.3% of 1920, still within 18–68% safe zone (960px high). That is why Shorts with 18 words in 4s get critical — the math predicts missed punchlines and the safe zone predicts cropped lines.

Test on the device, not the monitor: export a 10-second 32-char 32px test, AirDrop to the phone, hold at 12 inches, read the smallest line. If you can read 10pt captions at 18–68% safe zone, the caption is ok. If 10pt is 3px and blurred, the caption is warning even though the math says ok — the math is for area, not font rendering, so also check safe zone.

For monetization, a critical caption that is too fast loses watch time — the viewer rewinds 15% of the time, which is why split-screen and caption both matter for retention, not just composition.

Keep the caption consistent across episodes: a show that is 12 words for 4s every week builds a visual language, 18 words one week and 12 next breaks it. Log the WPM per episode and keep it ok.

For preservation, keep the native caption file at 32 chars/line and 32px — the native is the truth you can reconvert later with better framing.

Keep one spare caption reference file: a 10-second 32-char 32px caption at 12 words in 4s should read 180 WPM ok — drop it quarterly to verify the math still holds after browser changes. If it reads warning, the math drifted.

For a11y, keep WPM ≤180 and chars ≤32 and size ≥32 for every deliverable — the phone keeps the story where you left it.

Finally, use per-client caption sheet: Client | Words | Duration | WPM | Verdict — after 8 clients you see which client’s script is always 270 and you preset 12 words for that client, not 18.

For a11y, keep captions as sidecar SRT when possible — captions follow video time, not WPM, so WPM does not move captions, but a critical WPM with missed words will make sidecar captions hard to follow — another reason to keep ok.

Keep one spare caption reference in the vault — re-measure after every browser update; if it drifts, recalibrate.

Finally, keep WPM ≤180 for every master — the story stays where you left it on every display.

Font size and safe zone in practice

On 1080×1920, 32px is 96px high for two lines at 32px each plus 8px leading — 4.9% of 1920, well within 18–68% safe zone (960px tall, 345–1305px). At 22px, two lines are 44px high — 2.2% of 1920, still within safe zone but illegible. The safe zone does not save small font — 22px at 50 chars is still critical for size even though it fits in safe zone, because size is absolute, not zone.

Test on the device: export a 10-second 32-char 32px test, AirDrop to the phone, hold at 12 inches, read the smallest line in the safe zone. If you can read 10pt captions at 18–68%, the caption is ok. If 10pt is 3px and blurred, the caption is warning even though the math says ok — the math is for area, not font rendering, so also check safe zone.

For a11y, keep captions as sidecar SRT when possible — captions follow video time, not WPM, so WPM does not move captions, but a critical WPM with missed words will make sidecar captions hard to follow — another reason to keep ok.

Keep one spare caption reference file: a 10-second 32-char 32px caption at 12 words in 4s should read 180 WPM ok — drop it quarterly to verify the math still holds after browser changes. If it reads warning, the math drifted.

Line breaks and hyphenation

A 42-char line at 32px on 1080×1920 is 42×18px per char + spacing ≈ 756px wide — fits with 162px margin each side. 50-char line is 900px wide — overflows. Split 50-char line into two lines of 25 at 32px — each 400px wide, both fit. The split adds one more line, but WPM unchanged — 10 words in 4s stays 150 ok, but CPS per line halves, so critical becomes ok.

Hyphenation at 32 chars should break at word boundary, not mid-word — “readability” split as “read-\nability” is critical for comprehension even though chars ok. Keep words whole when splitting.

Duration and WPM for different genres

Gaming Short 18 words in 4s at 270 critical is expected — gaming chat is fast, but the punchline at the end is missed. For gaming, 12 words in 4s at 180 ok is better — punchier, more replays. Sermon Short 14 words in 5s at 168 ok is already ok — sermons are slower, viewers expect 150-180, not 270. Keep sermon at ok, not warning, even if you can fit more words — the audience is listening, not scanning.

Troubleshooting — extended

Says ok but viewers say too fast — check CPS and word length. Says warning 190 but no complaints — 190 is warning 180-220 — 15% miss, not 35%. For a hook, tolerable; for punchline, need ok.

Says critical 270 but you like it fast — 270 is 4.5 wps — 40% miss. If the last 3 are hashtags, tolerable. If every word matters, need ok.

Keeping the bar — per-editor caption sheet

Log every video: Video | Words | Duration | WPM | Chars | Size | Verdict — after 10 you see which editor always writes 270 WPM and you brief to write 12 words for 4s, not 18. Use that sheet to keep ok.

Keep one spare caption style guide in the vault — re-measure after every editor change; if it drifts, recalibrate.

For preservation, keep the native caption file at 32 chars/line and 32px — the native is the truth you can reconvert later with better framing.

Keep one spare WPM reference file: a 10-second 12-word caption at 4s should read 180 WPM ok — drop it quarterly.

Finally, keep WPM ≤180 for every deliverable — the story stays where you left it on every display, and captions stay readable.

Use per-client caption sheet: Client | WPM | Verdict — after 8 clients you see which client’s script is always 270 and you preset 12 words for that client, not 18.

For field recorders, keep the caption at 32 chars/line and 32px if deliverable is 9:16 — captions are not free, readability is.

Keep one spare safe zone check: 32 chars at 32px in 18–68% should read ok — drop it quarterly.

Finally, keep caption readability ok for every master — the story stays where you left it on every phone.

For a11y, keep captions as sidecar SRT when possible — captions follow video time, not WPM, so WPM does not move captions, but a critical WPM with missed words will make sidecar captions hard to follow. Keep one spare caption reference file: a 10-second 32-char 32px caption at 12 words in 4s should read 180 WPM ok — drop it quarterly to verify the math still holds after browser changes. If it reads warning, the math drifted. Finally, keep WPM ≤180 for every master — the story stays where you left it on every display.