Signal = The digital data as it was stored in its original form either on the CD or file or wherever you have it, currently being transferred over the network.
The key question is A) are your devices lowering noise that is generated from other devices in your house and/or external sources like from the power grid etc? Or B) are they removing noise that is present in the original data, aka your CD or digital file?
If the answer is A, then hopefully you will agree that the best you can hope for even theoretically, is that the signal remains untouched = Not degraded, but also not improved. It can however be improved when compared to another setup. But not improved when compared to the original source data.
If the answer is B, then I am at a loss, as I do not understand how these devices can improve upon the inherent sound quality of the original data.
The actual answer is C….they are removing and/or not adding noise caused by actually obtaining, transporting and playing the file.
Let’s say you start with both a flawed and a ‘perfect’ file. In either case, obtaining (reading a disc or delivery by internet) then moving the files across a network adds noise in a variety of different physical forms. In a network the data stream is regenerated multiple times, so first thing, there’s no such thing as the ‘original’. The original may be converted to a series of voltage based polarity switches, modulated by time, it may be converted to radio waves, modulated by time, light waves modulated by time Etc. At every stage, noise is added. The flawed file becomes more flawed and the perfect file becomes less perfect. When you play both, both have the potential to sound better, because the noise that was added by the entire replay process can be reduced or removed to a greater or lesser degree depending on the network design.
So how do you improve the original file? Each time the file is converted and regenerated it takes on the physical attributes of what does the conversion. If for example you transmit a file from a router using a 100ppm accurate oscillator (clock) the file includes a certain statistical timing error rate given by the oscillator. If you then buffer the file then retransmit it using a 3ppb accurate oscillator, you have essentially rebuilt both the voltage layer and the timing layer of the file. You have statistically a much lower timing error rate. The file is no longer the original…its a new version, maybe in the same or a different form; light, or radio frequency or voltage.
Again if you obtain the original file from a router, the voltage layer of that file is built using the router’s $7 switch mode power supply. If you then passed that file through a built-for-audio retimer/switch powered by a high quality low impedance linear power supply, your original file now has less noise, courtesy of the noise reduction circuitry, more accurate/faster polarity switches (higher quality square waves) thanks to the low noise, low impedance supply and a far more accurate time base, thanks to a better oscillator with better timing stats and less PS noise. .
If at each step of the process you improve the specification of the physical layer used to regenerate the file, you improve the physical quality of the file. A file with perfect physical quality i.e no noise, sounds better than a file with lots of noise.
There are lots of areas that can be improved. I’ve mentioned timing, by virtue of the oscillators used. There are power supplies, that can have less noise and lower impedances. EMI can be avoided and removed. Cables can screen better, components and connectors can be vibrated less. Network traffic can be reduced, latency can be reduced, the need for error correction can be reduced, the list goes on
The bottom line is that a digital file that has less deviations of any sort from the ideal physical layer specification sounds better. Those deviations make absolutely no difference in an IT world. It’s only when you convert the file to a series of sound waves and evaluate them subjectively in the brain that you hear the characteristics that these imperfections impart.
Most likely those imperfections impair the brain‘s ability to differentiate very fine differences, meaning that a. it’s working a lot harder to understand and interpret what’s going on and b. Some subtleties and details are not heard at all.
The bit pattern of the file doesn’t need to change, it just needs to have its inherent physical attributes improved. IT doesn’t give a damn about such things, but IT isn’t trying to interpret a series of resulting sound waves to impart musical sense, and meaning. The easier and more effortlessly the brain can do that job the more you’ll enjoy the music.
IT works just as well on a noisy file as it does on a perfect file, so if you apply IT related standards to audio files you end up with a file that’s as good as necessary but no better. Due to audio’s extra steps of conversion to sound waves and subjective interpretation by the brain, noise with and within the files matters as it has an impact on perceived sound quality. That’s the difference between processing files for IT and files for Audio. Audio is far more demanding of physical layer properties and attributes.
We both are fully aware in the real world mature adults contribute the lions share to what defines enjoyable youthful encounters. My comment was on the framework lacking completion of this timeworn custom. In the manner it was expected to occur. If not where it was expected to occur. 
