I’ve been experimenting some with the new Sample Sound Frequencies node in Blender 5.2. It’s so fun! I used it to create this audio visualization on my second channel.
I initially created it all in Cycles but then decided it was probably simple enough that I could do it in Eevee, so the final video is all Eevee (which was about 3x faster render times).
Total render time was about 15 hours. Which sounds a bit crazy for Eevee but most of the time was actually my PC sampling the audio over 7,500 times each frame.
I also realized that my idea of doing logarithmic frequency progression was a bad idea. Most of the lower tones aren’t very loud so it looks kinda lopsided. The frequency range was also between 10 and 10,000Hz which I feel like may have also been a bad choice. Of course, part of the issue is also that some of my earlier music had almost no bass sounds. That really didn’t help the visualization at all. 
The camera movement fades from manual to fully procedural after a few seconds (yeah I’m not animating 10 minutes of basic camera movement lol).
This is the GN set up. It’s a bit of a mess but I’ll share it anyway. 
Anyway, I learned a lot and I think I could probably make the GN node set up much better performance-wise so that it doesn’t have to sample the audio so many times. But it was good enough for my video so I’m done with the project for now.