The whole capacitor thing is interesting. Back in the day I developed a supermarket security gate to detect product tags and hence shoplifting, which for a few years was in use globally. This used two audio frequencies superimposed on a 20Hz triangle wave. The amp was class D (quite something for thirty-odd years ago) and hence needed an output filter. Designed one, and bought the bits - chunky inductor, and polypropylene capacitors. First thing was that the inductor melted, and that was when I found out about proximity effect. The second thing was that the capacitors howled like a banshee being tortured with a hot poker.
Anyway the capacitor thing was really interesting. I tried a variety of manufacturers, and found that acoustic output was determined by tight winding of the foils. Second that axial tubular ones were quieter than radials. That was down to the manufacturing process for the radials. First the foil is wound on a cylindrical former. The former is pulled out and what remains is squashed flat. The voids that are inevitably left cause the noise through electrostatic forces.
The quietest were cylindrical audio capacitors - a design now made by Kimber. Far too expensive for the product of course. But I found Roderstein axial ones that were also silent and cost much less. Roderstein is now owned by Vishay.
Anyway, it was an interesting exercise. As far as I know no-one tests audio grade caps by passing an audio band signal through them (of an amp or so) and listening to them. Any acoustic output is (a) frequency dependent and associated with mechanical resonance in the capacitor structure and (b) is clearly associated with a loss mechanism.
Aha - found a few left over - they were MKP1845, and Vishay still make them https://www.vishay.com/docs/26023/mkp1845.pdf . Just E6 values, which is a pain.