But the metanormalizer normalizes clips relatively to others, not independantly. Hence that does not fit the bill.
This being said, from the audio point of view, the original idea does not feel very good imho. For instance, what about the transitions between the normalized parts and the rest?
You should rather try a loudness maximizer, aka peak limiter.