Applies to: Words appear early or late relative to the associated speech or action.
The useful distinctions
- Captions are intended to relate to the programme’s audio.
- Edition identity matters when comparing text and media.
- One observed difference does not establish a constant offset throughout the programme.
Anchor the observation to content
Early and late describe a relationship between the text and a particular spoken or meaningful sound event. State which event the viewer used as a reference and identify the programme edition. A loose report that captions are five seconds out can lose the difference between a genuine repeated timing pattern and text that belongs to another cut. The viewer should preserve enough context for support to locate the same moment.
Keep the estimate separate from the pattern
An approximate interval at one point is useful, but it does not establish that every later caption needs the same adjustment. Record a second bounded observation if the problem persists and that check is practical. If the difference changes or begins after an edit, say so. Avoid altering files or recommending a universal correction from a single estimate. The immediate task is to describe the relationship accurately, including uncertainty about its scale.
Hypothetical worked example
For invented edition TM-5, a viewer estimates that text follows a spoken introduction by about two seconds at 03:10. Near 21:00 the relationship appears different. Their report preserves both observations instead of announcing a fixed two-second programme-wide delay. They ask support to confirm the corresponding text edition and review the cited moments. The example does not claim a hands-on test or prescribe any particular player’s timing controls.
A focused way to proceed
- Record the edition and selected text label.
- Identify one reference moment and the approximate early or late interval.
- Check whether a later comparable moment shows the same pattern before generalising.
When to stop or ask for help
Send the exact edition and bounded timing observations; ask about correspondence before applying a global adjustment.