I first tried reducing the volume of the clapping by -4dB, which made it more level throughout. However when I normalized to peak level in each channel it didn't do too much.
Then I tried normalizing via RMS, which kept the loud parts loud but brought up the lower levels which made it more level again. I know RMS normalization falls out of favor with a lot of people, so you might want to try something else, like fading out the crowd noise after each song, and then normalizing to peak level. Hope this helps.