Messy topic; polyester capacitors range from OK to pretty bad, depending on construction and materials. Also, sometimes the electrolytic can have euphonic benefits. I once changed an amplifier's input capacitor, (Citation 12 iirc) an electrolytic cap, to a well-reviewed polyprope. Change was a disaster. The softness of the electrolytic had masked the amplifier's typical early-transistor-design sound.
I believe in capacitor breakin. I find that a good rule of thumb is 40 hours. Ouch. Try not to listen during breakin, but just beginning and end.
It also depends on the drivers. A Scanspeak 8636 coated Kevlar mid and large Focal alnico tweeter in the '80's responded only mildly to improved resistors (Sandcast --> Mills) and minimal metallized polypropylene ---> better polyprope/foil. More recent project, using ATI 4" mid/Scanspeak 99000Revelator tweeter is sensitive to everything, as were the Accuton ceramics we also tried. The later drivers are also much more revealing with music.
Poke around on the web, and you'll find paraphrases of the Wireless World article; to the author's surprise, distortion was, in some cases, quite easily measurable for some capacitors. The earlierJung/Marsh article in Audio was the first to find a correlation between dielectric absorption and listening results, and also found that some types had measurable distortion.