Make the narrated passage match the passage being read

Bind the recording to the actual text revision

List the approved manuscript, recorded performance and EPUB export that are meant to work together. Identify passages changed after recording, including small qualifications, numbers and product names. A recording filename that contains the current campaign name does not prove it reflects the current words. Ask the editorial owner which differences are intentional adaptations and which require correction before synchronization begins.

W3C EPUB 3.3 describes media overlays as a SMIL-based connection between structured narration and corresponding publication content. This is more than embedding a sound file beside a chapter. The practical acceptance object is the relationship among a passage, its recording segment and its place in the reading sequence.

Choose boundaries that reveal a mismatched relationship

Create a compact test selection covering the beginning of the publication, a midchapter passage, a chapter transition and a recently changed section. Include a passage where a nearby heading or caption could plausibly be mistaken for the intended text. Record the expected spoken content and reading location in the worksheet so the reviewer does not simply follow whichever location the player happens to highlight.

Ask the production specialist to map any discrepancy to the text reference, recording segment or exported relationship. A late highlight may have a different cause from a sentence that was never recorded. Preserve those distinctions instead of applying a timing adjustment to every defect. Do not invent a universal millisecond tolerance; agree acceptance with the relevant reader experience and production requirements.

Inspect the listening route in named reading systems

On each agreed reading system, start from the selected passage and compare what is spoken with the text being followed. Pause and continue, move to another tested location and inspect the next transition. Record whether a reviewer can tell where they are without relying on memory of the manuscript. Use the final packaged candidate rather than an isolated audio editor preview.

The W3C reading-system specification recommends media-overlay support for systems capable of prerecorded audio; support is not universal. Separate an unsupported feature from a defective implementation in a supported system. Have the release owner describe the observed scope accurately, and verify the ordinary reading route independently rather than claiming that a silent device necessarily contains a broken recording.

Close defects at the relationship, not only the file

For each defect, retain the passage identity, observed speech, expected relationship, reading-system version and candidate export. Route editorial mismatches to the editor and mapping or packaging defects to production. After a repair, repeat the affected passage and its neighboring transitions. A corrected recording can still be paired with an obsolete boundary map, so replacing one asset is an intermediate step rather than closure.

Approve the coherent text, audio and mapping revision together. Record remaining limitations in language the publishing owner can use without promising support everywhere. When text is reordered or audio rerecorded, reopen the affected synchronization evidence even if the publication title is unchanged. The record should establish that the intended reading path was demonstrated, not imply that a short sample proves every passage or every assistive technology works.

Make the reasoning concrete.

A fictional craft publisher revises a project guide from six paper sheets to eight after the narration is recorded. The EPUB displays the updated paragraph, but its synchronized clip still reads the earlier material list. Both the audio file and the visual page passed separate reviews.

The publisher holds the paired passage until its text and narration agree. It preserves separate audio-quality checks while adding relationship-level acceptance to the publication release.

Take the questions into your next brief.

Use these prompts in your own notes. Nothing is submitted from this page.

1. Revision

   Which text, recording and export are intended to be paired?

2. Passages

   Which changed or boundary passages are most revealing?

3. Observation

   What speech and reading location occur together?

4. Scope

   Which reading systems support and demonstrate the intended route?

5. Repair

   Was the corrected relationship retested with adjacent transitions?

Watch for the wrong conclusion.

• Approving clean audio without comparing its manuscript revision.
• Treating every silent reading system as a corrupt audio file.
• Replacing a clip without retesting its associated passage.

Sources & scope

Workflows, worksheets and fictional examples are original editorial analysis. W3C specifications support only the attributed format and reading-system distinctions. Named-version tests, qualified accessibility review and editorial acceptance remain separate; no universal compatibility or conformance claim is made.

• W3C: EPUB 3.3 →Primary W3C technical specification · Supports: Supports the text/narration synchronization purpose of media overlays.
• W3C: EPUB Reading Systems 3.3 →Primary W3C technical specification · Supports: Supports bounded reading-system media-overlay support.