Summary: Speaker-identification dialog re-prompts every time the project is reopened, even though the relevant content is cut, muted, and excluded from the script.

Details: I have a composition built from a sequence combining two recordings — a Speaker track and a Slides track. The Slides track's audio is muted and it's excluded from the script. I also cut the opening ~24 seconds (a countdown clock) from the beginning of the composition. Despite this, the speaker-identification dialog tries to identify a voice in that same cut/muted/script-excluded opening section — and it does this every single time I reopen the project, not just once.

I raised this in today's community office hours (Sept 23), and was told that repeated re-prompting on every project open "seems not right" and wasn't something they could explain. I'm filing this as the specific bug: the one-time scan-the-source behavior may be intentional, but it shouldn't need to repeat on every reopen once already dismissed/resolved.

Steps to reproduce:

Create a sequence from two recordings (Speaker + Slides).

Mute the Slides track's audio and exclude it from the script.

Cut the opening portion of the composition (before the actual content starts).

Create a composition from the sequence.

Close and reopen the project.

Speaker-ID dialog appears, referencing content in the cut/muted/excluded section.

Repeat steps 5–6 — the dialog reappears every time.