How would you review faint digital noise before mixing?

I help a community archive prepare old concert recordings for private sharing with former performers. One recording has a lovely room sound and an expressive singer, yet several quiet passages contain small digital chirps that pull attention away from the music. I want to improve listenability without polishing away the atmosphere that makes the recording meaningful. During my research I found a brief repair guide. It seems relevant, although I would rather hear from people who have restored imperfect live material. My question is how to set a sensible stopping point. The historic ambience is valuable. A brief sample can guide the first decision. I can compare matching passages before processing the full file. The restoration should remain reversible. Any change must respect phrasing. I need documented settings for future copies. A short listening pass may expose unwanted side effects. The audience reactions should stay present. The singer's voice must remain direct. Each adjustment should remain clear to document. I will compare it at quiet volume. I would rather preserve a little imperfection than make the result sterile. How do you decide when a restoration has become too intrusive? I am grateful for thoughtful suggestions.