Did you try Chris’s Compressor? Chris designed it so he could listen to wildly changing volume opera in a noisy car. I use it at a stiff correction setting for a favorite spoken word podcast that has no controls at all, but it may be perfect for you at its default setting.
Most of the volume setting tools start the day with the wrong assumption. They work on volume peaks and momentary loud events. That has only limited presentation usefulness, but it’s insanely easy to program. Much more difficult is perception and loudness. That’s not simple maths any more, now you have to model your ear.
That’s what Chris does.
https://theaudacitytopodcast.com/chriss-dynamic-compressor-plugin-for-audacity/
You can also use the combination of RMS Normalize (a new tool from Steve) and Effect > Limiter. That’s 2/3 of the tools used to tame audiobook readings.
https://forum.audacityteam.org/t/rms-normalize/45334/1
Used in this case at default -18dB and Limiter to -1dB or -2dB. Limiter -3.5dB is the audiobook standard which stiffer than you probably need.
Koz