How to Set an Accessible Caption Reading Speed

Reading speed is one part of caption quality

A caption can meet a numerical speed target and still be inaccessible because it is inaccurate, late, incomplete, poorly placed, or split in a way that changes meaning. Conversely, a cue slightly above a preferred target may be the least harmful option when shortening it would remove essential information or extending it would reveal dialogue too early.

Use reading speed as a controlled editing constraint inside this larger quality review. The objective is not a green calculator result; it is a caption viewers can read while following the program.

What WCAG requires—and what it does not specify

WCAG 2.2 Success Criterion 1.2.2 requires captions for prerecorded audio content in synchronized media at Level A, with a defined exception when the media is itself an alternative for text and is clearly labeled. Success Criterion 1.2.4 requires captions for live audio content in synchronized media at Level AA.

WCAG defines captions as a synchronized visual or text alternative for speech and non-speech audio information needed to understand the media. Captions therefore include relevant speaker identification, sound effects, music information, and other necessary audio—not dialogue alone.

Calculate CPS and WPM from the cue duration

Characters per second divides a cue’s character count by the time it remains visible. Words per minute divides the word count by the duration in minutes. CPS is often more stable for timing comparisons because captions with the same word count can contain very different word lengths.

For example, a caption containing 68 normalized characters displayed for four seconds runs at 17 CPS. If it contains 12 words, its secondary estimate is 180 WPM.

Counting conventions differ between tools and specifications. The DOFIW calculator normalizes repeated whitespace and line breaks to single spaces for its CPS count, while preserving original line breaks for line-length checks. Document the convention when comparing results from different systems.

Choose a target for the project, not from habit

The calculator offers adjustable starting profiles: 13 CPS for an easy pace, 17 CPS for a general pace, and 20 CPS as a fast upper editing target. These values let editors compare a cue against a declared workflow. They are not universal claims about what every viewer can read.

Choose or customize the target based on:

• The receiving platform’s current specification and counting method.
• Whether the content is for children, emerging readers, language learners, or a general adult audience.
• The complexity of vocabulary, numbers, names, instructions, and unfamiliar concepts.
• Whether viewers must also study charts, demonstrations, signs, or other visual detail.
• The language and writing system of the captions.
• Whether captions are same-language, translated subtitles, live captions, or edited prerecorded captions.
• Viewer testing and the organization’s documented accessibility standard.

Use a slower target when the audience or visual task needs more processing time. Do not assume a fast-speaking source automatically justifies unreadably fast captions.

Fit reading time inside accurate synchronization

A cue should appear when the corresponding speech or meaningful sound begins and disappear when it ends or at an editorially justified boundary. Captions that arrive early can reveal information or punchlines. Captions that lag require viewers to remember audio or visual context while reading delayed text.

The calculator’s suggested duration is mathematical: normalized character count divided by the chosen CPS. It does not know speech boundaries, shot changes, adjacent cues, minimum or maximum durations, gaps, overlaps, or dramatic timing. Use the number to identify a conflict, then solve the conflict while watching the video.

Repair a caption that is too fast

Use the least damaging option. The right repair depends on available silence, shot boundaries, dialogue rhythm, surrounding cues, and the importance of every word.

1. Verify the timecodes.Correct accidental early out-times, late in-times, frame conversion errors, gaps, or overlaps before rewriting text.
2. Extend within the matching audio context.Borrow safe time before or after the cue only when synchronization and adjacent captions remain clear.
3. Merge compatible cues.A combined cue may create a steadier reading rhythm when the same speaker and thought continue naturally.
4. Split at a natural boundary.Use clauses or complete phrases, and give each cue enough time to be noticed and read.
5. Remove verbal clutter carefully.Delete repetitions or fillers only when they carry no meaning, characterization, tone, hesitation, or plot function.
6. Rewrite for clarity when permitted.Use concise syntax without changing facts, intent, humor, register, or the speaker’s identity.
7. Escalate the conflict.If platform rules, required verbatim text, and available timing cannot all be satisfied, document the issue for editorial direction.

The shorter version preserves the instruction and sequence. A dramatic line, legal quotation, lyric, or character-defining hesitation may not permit the same degree of condensation.

Break lines by meaning, not character count alone

A cue can be within its CPS target but still difficult to read because a line separates words that belong together. Keep articles with nouns, prepositions with their phrases, names together, auxiliaries with main verbs, and short subject-verb units intact where space permits.

The tool’s preferred line length defaults to 42 characters and flags more than two lines. Both are editable review targets, not universal WCAG limits. A platform may specify a different safe area, line count, typeface, or character limit.

Do not create three rapid cues only to avoid a two-line caption. Viewers need time to notice each change. Evaluate cue duration, number of visual updates, and phrase integrity together.

Count speaker labels and meaningful sounds as real caption content

Speaker identification, sound effects, music cues, and relevant vocal information consume display space because viewers must read them. Include them in the speed calculation and do not remove them merely to lower CPS.

• Identify a speaker when the visual context does not make the speaker clear.
• Describe a sound when it contributes information, mood, location, timing, or plot.
• Distinguish sound from visible action: caption what is heard, not everything seen.
• Use concise, consistent labels such as [door closes softly] or a project-approved style.
• Preserve relevant hesitation, emphasis, laughter, tone, and music information according to the content and style guide.

A cue containing only [music] may be too vague when the type, source, or change in music matters. A highly detailed label can also become distracting. Describe the audio information needed to understand the scene.

Review placement, contrast, and competing visual tasks

Captions should not cover names, translations, demonstrations, scores, diagrams, or other relevant visual information. Move captions consistently within the platform’s safe area when necessary, and ensure they do not overlap one another or run off screen.

Caption text and its background or edge treatment need sufficient visual contrast. Test light and dark scenes, gradients, motion, and user-adjustable styling. Closed-caption systems may let viewers select size and colors; avoid authoring choices that prevent user preferences from working.

A mathematical target cannot account for this competition. Use a slower pace, clearer script, repeated visual information, or more production time when the scene requires simultaneous reading and observation.

Review live captions differently from prerecorded captions

Live captions face unavoidable latency and fewer opportunities for offline correction. Reading speed still matters, but quality review must also examine accuracy, delay, dropped content, speaker changes, punctuation, and whether captions continue through the entire event.

Do not evaluate a live stream only by sampling a quiet minute. Include fast exchanges, names, numbers, audience questions, technical terms, music, remote speakers, and transitions. Provide captioners with preparation materials and a list of names and vocabulary. Re-edit captions before publishing a recording when practical.

Test captions in the final WordPress media player

A correct WebVTT or other supported caption file can still fail in delivery if the track is not attached, language metadata is wrong, the player hides controls, or theme styles make text unreadable. Inspect the published page.

Test the hosted file and player combination you actually publish. Browser, platform, and player support can differ from an offline caption editor.

What the Caption Reading Speed Calculator can—and cannot—do

The DOFIW Caption Reading Speed Calculator processes one pasted cue locally in the browser. Caption text and timecodes are not uploaded or sent to an API. It reports duration, normalized characters, words, CPS, WPM, line count, longest line, minimum duration at the selected target, and a suggested end time.

The tool does not parse or modify SRT, WebVTT, or other subtitle files. Its WPM count assumes whitespace-separated words, so it is not suitable for every language. A passing result does not establish WCAG conformance or legal compliance.

Use a complete caption review workflow

1. Confirm the governing specification.Record platform, language, audience, format, CPS method, line rules, and any regulatory requirements.
2. Check text accuracy and completeness.Compare dialogue, speakers, sounds, music, numbers, names, and punctuation with the audio.
3. Check in-times and out-times.Align cues with the corresponding audio and preserve dramatic or instructional sequence.
4. Calculate speed.Find dense cues and compare them with the documented target and counting convention.
5. Repair in context.Extend, merge, split, or condense without losing essential meaning.
6. Review line treatment.Keep natural phrases together and respect the delivery safe area.
7. Watch the full program.Evaluate rhythm, placement, scene changes, competing visuals, and cumulative fatigue.
8. Test the published player.Use keyboard, zoom, mobile, full screen, seeking, and playback-rate controls.
9. Include representative viewers.Use feedback from deaf and hard-of-hearing people and others in the intended audience.

Common caption-speed mistakes and direct fixes

• Treating one CPS value as law: follow the applicable platform or project specification and document the target.
• Using WPM alone: compare CPS as well because word lengths vary.
• Extending through unrelated speech: repair timing without breaking synchronization.
• Deleting meaningful sounds: preserve audio information required to understand the content.
• Condensing away tone or intent: verify meaning, characterization, humor, and register.
• Splitting at the character limit: move the break to a natural phrase boundary.
• Creating rapid caption flicker: balance line length with sufficient cue duration and stable segmentation.
• Ignoring speaker labels in the count: calculate every character viewers must read.
• Checking cues in isolation only: review adjacent captions, shot changes, and the full program rhythm.
• Assuming an accurate transcript is enough: synchronized video requires captions under the applicable WCAG criteria.
• Testing only in the caption editor: verify the final WordPress player and responsive layout.
• Calling a passing calculation accessible: review accuracy, completeness, placement, controls, and viewers’ experience.

Caption reading-speed checklist

Continue the DOFIW caption workflow

For wording decisions, use How to Improve Web Readability with Plain Language.

Frequently asked questions