I haven't seen any replies about this, so I'd like to bring some attention to two characteristics of capacitors that actually can affect their audible and electrical performance.
1. The first is actual impedance versus theoretical impedance. Any passive or active component only operates properly within its expected frequency range, and if it is used at frequencies beyond that range they won't perform in the expected way. RF engineers are painfully aware of the fact that standard diodes and transistors just don't work when you get into the microwave range. Both resistors and capacitors however, can exhibit undesired performance at an eventual high frequency too, where their parasitic inductance becomes significant. Since a lot of capacitors are actually rolled up like cinnamon rolls to cram the two plates into that small space, there is indeed a frequency where the parasitic inductance takes over. If you plot the measured impedance of say, a large electrolytic capacitor, you will eventually find there is a frequency where this inductance kicks in, and it will look more like an inductor instead of a cap. Hopefully that happens well above the audio range so it doesn't screw with your beautifully designed crossover networks, and amplifier feedback loops. This is the real reason that some folks put small value parallel capacitors across those big electrolytic caps, and the rule of thumb often used is to choose a value that is 1/20th the value of the big cap. If you want to get it just right, go ahead and measure the actual impedance performance of those big electrolytic caps in the power supply and those used as coupling caps in the signal path. and choose the correct cap to put in parallel to allow the impedance to continue to drop like it's supposed to.
2. The other capacitor defects are dialectric absorption (DA), and dissipation factor. In simple terms, DA is when a cap that had a charge on it is expected to allow you to suddenly remove that charge. This happens thousands of time a second with audio signals, but caps that aren't so good can have a tendency to hang onto that charge instead of letting it go. This can be audible if it is significant enough, and you will find that very high quality caps will have low DA. Now its quite possible that some guitar amp users may like this sloppy cap performance and that's fine if it make the sound you're looking for. Measuring DA with actual signal sources is not often discussed, but there was an article in Audio Amateur issue 4/1985 page 22, that describes a circuit that you can build to see the effect right on your oscilloscope. I believe there is still a copyright holder on the articles from Audio Amateur, so I can't reprint it, but I will show you the schematic and photos of the one I made, and oscilloscope pictures of some of my measurements using it. Interestingly, John Roberts, one of our regular forum contributors is mentioned at the end of the article as someone who wrote a letter to Audio Amateur about this test.
This circuit uses a reference cap, and has a bridge network that allows you to compare a second cap to it, and display the residue difference on the oscilloscope. I used a Kemet metalized polypropylene 1 uF/630V cap for the reference, and compared it to an old tubular Wima, a new stacked Wima MKS, and finally an old Sprague tubular, probably mylar. The two Wimas are almost identical, but the Sprague has MUCH more residue, especially when you notice that the 'scope's vertical sensitivity is set at 8V/div instead of 500 mV/div. The signal source is a pulse generator with a pulse width of 10 ms, and a period of 30 ms. The source amplitude is about 12Vpp.
When I first built it, I used tiny trimmers for the pots, but that made it really annoying to use, so this version used regular pots and knobs, which made it much easier to adjust. The Radj2 is to compensate for the ESR, while the other 3 pots adjust for the time constant. C1 band limits the input signal. The differential instrumentation amp with precision gain shifting is pretty expensive at $59.29 from Mouser, but the rest of the parts are pretty cheap, and it makes a hell of a cap tester. It will tell you which caps are really really good, but it will also tell you what to look for in a weird cap that you have decided you really like in your guitar amp. Have fun.
The things that make the difference IMHO in a specific amp's sound are valves/tubes, bias, and of course transformers. Of course different caps may change things, but the change might not be an improvement. I had people coming back complaining about changed sound after repairs to factory spec, but usually it was that the valves or transformers had been changed and the bias put to where it should be as per spec not where it had drifted to after being on the road. Consistent valve amp sound is hard. Transistor amps were much easier. Now I use modelling to get where I want to go! 
