The production path
A regeneration request identifies a stored section and creates a replacement through the same selected voice workflow. After the provider returns audio, the application validates the replacement, re-merges the complete take, rebuilds the SRT, and promotes the new revision only after the publish checks pass. The replacement section generates on the exact same quality lane and Credit multiplier as the rest of the take — one of the seven lanes from Core through Apex — so a regenerated section never quietly changes cost or quality tier.
- The original published take stays available if generation, merge, storage, or validation fails.
- Revision keys and job guards reduce duplicate or concurrent promotions.
- The regenerated section is highlighted in the rebuilt SRT proof so the change is inspectable.
What it does not promise
The current capability is section-level, not arbitrary sentence-level replacement. A section can contain more than one sentence, and the rebuilt timing can change for the surrounding cues. This boundary is intentional: it keeps the workflow explainable and safer to roll back.
When to use it
Use regeneration for factual corrections, revised openings, changed calls to action, or a pronunciation fix. Listen to the transition into and out of the replacement, then download both artifacts from the new revision.