Caption Quality Self-Audit Checklist (PDF)

Introduction

Getting captions running is the first milestone — but running captions and accurate captions aren't the same thing. The widely recognized industry standard, set by the Described and Captioned Media Program (DCMP) and echoed in FCC caption quality guidance, calls for 99% accuracy — roughly one error allowed per 100 words. Automated captioning systems can approach that benchmark once custom vocabulary is configured. However, errors still slip through: misheard names, garbled technical terms, missed speaker changes, or captions that lag behind the speaker.

This checklist gives you a repeatable way to spot-check caption quality after a meeting, so problems get caught and fixed before residents rely on inaccurate captions — or before an error ends up in your permanent archive.

Use it after any live meeting, or when reviewing captions on recorded and archived content. It takes about 10–15 minutes per meeting once your team is familiar with the process.

This resource is for educational purposes only. It supports your video accessibility efforts and helps your agency work toward WCAG 2.1 Level AA. It is not legal advice, and completion does not guarantee a specific legal or regulatory outcome. Consult qualified legal counsel for guidance specific to your agency's situation.

A printable PDF version of this checklist is available for download here.

WCAG and FCC: two different standards

These two frameworks get blended together often enough that it's worth separating them clearly. WCAG 2.1 — specifically Success Criteria 1.2.2 (Captions, Prerecorded) and 1.2.4 (Captions, Live) — is the accessibility standard behind your agency's legal obligations under ADA Title II. WCAG requires that captions be provided. It does not itself specify a numeric accuracy percentage.

The 99% figure comes from a different place: the FCC's caption quality rules (47 CFR § 79.1), written for broadcast television, along with industry guidance from groups like the Described and Captioned Media Program (DCMP). Neither of those directly governs government web video the way WCAG does. But their accuracy/synchronization/completeness/placement framework is the most widely recognized, practical definition of what "good captions" look like — which is why this checklist uses it as the quality yardstick. Meeting that bar will satisfy your WCAG obligations; it's just not the same rule imposing the requirement.

What good captions look like

Caption quality is generally measured against five standards:

• Accuracy — Captions correctly represent what was said, with 99% or higher accuracy under normal speaking conditions — the DCMP/FCC industry standard (supports WCAG Success Criterion 1.2.2 and 1.2.4)

• Synchronization — Captions appear within about 3 seconds of the spoken words, so viewers aren't reading text that's noticeably out of sync

• Completeness — All spoken content and relevant non-speech sounds (applause, alarms, crosstalk) are captioned, not just the "important" parts

• Placement — Captions don't obscure essential visual content, like a speaker's face, a slide, or a vote tally on screen

• Speaker identification — When multiple people are speaking, especially during debate or public comment, it's clear who is saying what

A caption feed can look fine at a glance and still fail one or more of these standards. The downloadable checklist above walks through each one step by step — sampling your content, checking accuracy, synchronization, completeness, placement, and speaker identification, then documenting what you find.

When to do a deeper review

A quick sample-based audit is enough for routine monitoring. Do a more thorough, full-length review when:

• You're launching captions for a new meeting type or content format for the first time

• You've made a significant change to your custom vocabulary or captioning configuration

• A resident or staff member reports a specific caption error

• The content covers a legally significant decision, such as a budget adoption or a contested vote, where accuracy matters most

MediaScribe integration

If you're using MediaScribe, the caption editing interface lets staff review and correct captions directly against the recorded audio, and any corrections you make can inform updates to your custom vocabulary list going forward. Caption coverage and accuracy patterns can also be tracked over time using exported reports, which is useful documentation for the "document and follow up" step in the checklist.

Summary

• Running captions and accurate captions are not the same thing — quality needs a periodic check, not just a one-time setup.

• Review captions against five standards: accuracy, synchronization, completeness, placement, and speaker identification.

• Sample a few representative clips rather than reviewing entire meetings; look for patterns, especially recurring vocabulary errors.

• Document what you find and fix recurring problems at the source, such as updating your custom vocabulary list.

• A quick self-audit takes about 10–15 minutes; do a deeper review for new content types, configuration changes, reported errors, or high-stakes decisions.

What's next

With your caption review process in place, the next step is making sure the rest of a meeting's presentation supports accessibility too — not just the captions. The Accessible Meeting Presentation Checklist walks through room setup, visual display placement, and presentation habits that help captions and audio descriptions do their job well from the start.

MediaScribe Academy · Video accessibility resources for government staff · Version 1.0 · January 2026