A generatively extended shot does not automatically guarantee a matching transcript for the added seconds. Inspect the real audio and the editor's current limitations before filling the caption gap. Adobe's Generative Extend known-issues page documents transcription-related limitations for extended material. This is a specific external-editor boundary that can change as the feature evolves.
Locate the added region
Keep the approved pre-extension version and identify where original material ends and generated material begins. Record the added time span. Inspect the waveform and listen to that region without assuming it contains new dialogue. An extended picture may contain silence, ambience or another audio result rather than a continuation of the speaker's words.
Compare the existing transcript with the actual speech. If the original speech was already transcribed correctly, preserve that approved text. Check the current feature documentation and your application version before assuming that another automatic transcription pass will process the extension.
Decide whether a caption belongs there
If no speech exists, do not invent spoken words simply to fill every second of the shot. If an accessibility caption for a meaningful sound is appropriate, describe only the verified sound under the project's caption policy. A visual scene continuing does not itself require a dialogue cue.
For a hypothetical family memory film, a shot may be extended to hold the image after the narrator finishes. The extra seconds may need no new dialogue captions. Another extension could contain audible material that requires manual review and transcription. These are assumed examples, not a promise that the generation preserves speech or produces usable new narration.
Where speech is present and understandable, transcribe it carefully and have the owner approve any substantive words. Where it is unclear or unintended, consider replacing or removing the audio through an approved edit rather than writing plausible dialogue. Keep generated words distinct from statements the real person actually recorded.
Rebuild timing and verify
Adjust cue timing to the real final audio. Check that the last original caption does not remain on screen across unrelated added material. Review any downstream timing changes caused by the extension, including the ending and music transition. A caption gap and a shifted later cue are separate checks.
Export the affected section and watch it with sound at normal speed. Then inspect the full delivery after all changes. Save the approved transcript and extension boundaries with the project. Acceptance means every caption reflects verified content and correct timing, including a deliberate absence of dialogue where appropriate. For Imagild Film Studio material edited with generative tools, review the extension as new material rather than assuming the previous transcript approval covers it.