Sorry about that. It seems like assumed too much knowledge on your part. It’s kinda hard to determine in what way to answer. When being too simple it sounds really condescending if you’re talking to someone with knowledge and also waste quite a bit of time.
But assuming too much leaves issues like this. Sorry for that.
Basically your second guess was pretty good. This is close to what I mean.
The boolean that you wanna feed into your branch is the result of a test with your vector length from the normal impulse.
You simply compare if it is higher than an arbitrary number which you will have to determine. This is why the print comes in handy. You can just throw down a ball and test at which point it starts to feel / sound weird if it still makes that noise. You can also decide to apply a different pitch or volume depending on how hard the impact is (which is your vector length. It’s a value which tells you with how much force the object hit the other object).
And like this you have to figure out which value fits best for your case. At which point shouldn’t it make a sound anymore? Should it affect pitch / volume? How so? etc. It’s a lot of testing what seems right and decisions which you will have to make.