Thank you @almarg. I was going to comment as I can accept disagreements, but I cannot accept false statement about what I say.
That being said, I also have experienced the 2-3db effect when changing fuses or power cords or op amps. I'm going out on a theory here and I'm sure there will be a lot of conjecture to this, but I believe this all has to do with how a particular circuit "slews" in response to the input waveform. This also has mostly to do with digital sources (if we played from an analog tape, we probably wouldn't have this scenario).
Waveforms from a DAC chip are always created by the DAC chip sending out flat-line DC at different voltage levels. With a 16-bit CD, we have the ability to define 65,536 different possible voltage levels between the lowest point on a waveform and the highest point. However, the output from the DAC chip is always DC (essentially a square-stepped waveform). If we put a tap on the output of the DAC chip and look at the total waveform, it looks like a true analog waveform to us. However, once you zoom into the waveform very closely, you will still see that it's just square-stepped DC at different levels. You might have to zoom 1000% or more to actually see this.
The fight we have with any audio chain based on a DAC is to smooth out this square-stepped waveform so that it resembles a true analog waveform (with no steps). This can be done by applying several/many op amp circuits. The negative feedback capacitor value can affect how this adjusts the waveform as well. Class A circuits are excellent for waveform "smoothing" as they have a very gradual response to the initial input and have more "decay" on the back side of the waveform.
When an op amp (or discrete circuit) responds to the square-stepped wave, it will try to "slew" up to proper voltage to generate this on the output. As soon as it starts "slewing" it will draw voltage from the capacitors in a power supply. Because the capacitor are "temporary" storage, the available voltage starts going down. This is where A/C will start to flow in and restore any voltage in the caps.
Now, if a circuit responds too fast to the incoming "square-stepped" waveform, then it will try to reproduce that "square-stepped" effect as accurately as possible. Slowing down the "response" to a change in input voltage is where want to go.
Different fuses can adjust things one way or the other. For example, using a silver-based fuse can allow very fast and very high conductivity when the op amp circuit tries to draw from the power supply. It will do as least amount of "waveform smoothing " as possible because it can respond and slew very quickly. However, putting in a gold-plated Isoclean fuse can slow down the "initial draw" of current from A/C. It's not the total amount of current we are limiting. It is the "response of the initial draw" (like a water faucet slowly turning on until it get to full output). The Furutech rhodium fuses behave in a different way. They are quicker responding than the gold-plated Isocleans, but not as quick responding as the HI-Fi Tuning silver. This slower response of "initial draw" can cause a more smoothing of the waveforms, causing us to get "louder noise".
It's all based on speaker drivers as well. The square stepped waveform (when measure by a scope) can seem to be correct size, but when most of this is very fast slews to the next point where we have an area of straight DC, you won't get the volume that you expect. Since what we hear as based on the speaker moving (i.e. response to "changes" in voltage), if there is any DC in the waveform, we will not hear it.
This also can explain why silver elements or op amps with extremely fast slew rates will tend to translate waveforms to be faster than what they really should be.
This is all theory based on what I have learned over the years, but I believe it can explain why one fuse will "sound louder" than a different fuse.
