I am aware that Playlist Rider is designed as a real-time component. However, since I often notice in the post-processing of dialog material that the gain riding seems disturbing / unnatural with eruptive speakers even at the Slow level, I would like to suggest an improvement.
Why don’t you follow Sibilance’s example and create a post-processing component and a live component?The post-processing component could then have a controllable lookahead of 0.5 to 5 seconds. This could significantly improve gain riding by allowing peaks and valleys within a possibly controllable dynamic range, or by controlling gain riding more smoothly around momentary peaks and valleys.
What do you think?