Has anyone built a lightweight pipeline to turn 30–40 minute docent tours into 45–60 second vertical stories that land with audiences? I’m piloting Descript + CapCut with auto-captions and a single narrative hook pulled from the first 10 minutes, and we’re seeing about 2.3x reach over static posts — how are you handling consent, on-gallery audio, and B‑roll rights?
We ask the docent for a 45–60s VO pickup right after the tour and cut to that, so in-gallery audio is just nat sound under it… Consent is a simple “recording in progress” sign with QR opt-out and blue-dot stickers — like a low‑key VIP badge — and B‑roll stays on a pre-cleared green list; anything extra is CC from https://openverse.org. That should slot into your Descript/CapCut flow.
Instead of a VO pickup like @cnelson55, we clip a tiny bodypack recorder (Instamic or RØDE) to the docent and duck gallery nat sound under it in the 45–60s cut; it solved on-floor audio without post headaches in Descript/CapCut. Consent is opt-out wristbands tied to ticketing, plus a no-capture note on the route, and our B‑roll pulls only from a pre-cleared “green list” baked into the tour script. Also don’t limit the hook to the first 10 minutes — tag laughs/gasps on the recorder and pull the strongest 6–8 seconds from anywhere in the 30–40.
Building on @cnelson55, skip re-records and mine the first 10 with Descript’s transcript; we pick one sentence, run CapCut’s noise reducer on just that clip, and time the one‑minute to it. Consent is a door placard — “filming for museum channels today” — with a QR to policy and “no photo” stickers at admissions; docents sign a season‑long release. For cutaways, we stick to Open Access works or architecture and swap in label/map panels for loans.