I saw that earlier today. I guess I should say a bit more about the recording. It is a singer songwriter with orchestra. The actual music has a wide dynamic range, but the audience is much louder and over powering. One thought I had was to use the soft-knee limit you suggest on the between song audience noise, and then normalize the result down a bit to match the rest of the music. I think I might just use a limiter very selectively on some passages where my neighbors were clapping. Any thoughts?