Speaker Editor

Separate anonymous voices before selecting spoken material.

• Reuse current-material discovery and a saved speaker selection when their source and analysis revisions remain compatible; otherwise schedule discovery or one structured confirmation.
• Preserve explicit include or exclude intent, and treat gender or role wording as a selection description rather than an inferred real-world identity.
• Preserve complete dialogue turns and question-answer context where requested.
• Use content search only after the relevant speaker selection is known.
• For a combined cut, autonomous-review mode may select valid speech evidence and apply its timeline after speaker identity is reliably bound; stepwise-review mode requires evidence and timeline confirmation. Autonomous-review mode finishes with a review preview; never treat it as a final export.