USB power from a computer is usually noisy. This is no problem for the digital data but SOMETIMES the noise leaks into the analog circuitry. It more often a problem with audio interfaces for
recording (the ADC part) because microphone signals are weak and the preamp boosts any noise along with the signal. USB power supply noise is usually a high-pitch whine.
because some people swear that the upgrade is transformative.
90% (or more) of the "audiophile community" is nuts!!! This site and HydrogenAudio are a couple of exceptions. One clue is the words they often use. If they say the sound is "more detailed" or "analytical" or a thousand other nonsense words, that's a big clue. ...Tell me if it has noise or distortion or something with a scientific or engineering definition.
Noise is a "big one" and noise is mostly an analog issue... MOST "defects" are analog issues unless you have "telephone quality" digital, or something like that.
Audiophoolery describes the 4
real characteristics that affect/describe audio quality... Noise, distortion, frequency response, and timing-related errors. The timing related stuff is mostly acoustical... i.e. Reverberation and standing waves and other effects of reflected sound in a room.
With
electronics, the main thing is
noise. Sometimes there is audible background noise (hum, hiss, or whine). Distortion and frequency response are
usually better than human hearing, unless you over-drive an amplifier into clipping (distortion) or something like that.
With speakers & headphones the "big thing" is
frequency response. Speakers have usually more distortion than electronics but it's usually not audible with music, again unless the speaker is over-driven.
Speakers or headphones (and room acoustics) make the biggest difference and every speaker sounds different, whereas "good" electronics neither degrades or (accidently) enhances the sound so there is usually little or no difference (except that one amplifier can be more powerful than another, etc.).
...Most of these things can be measured and quantified except with lossy digital compression (MP3 etc.) the noise & distortion changes from moment-to-moment depending on program content and that makes them difficult to measure. And it's often difficult to know when a measured difference makes an audible difference.
As a general rule, you can measure "better" than you can hear. Just for example, I have an oscilloscope on my bench at work that goes up to 100Mhz (that's "radio frequency" not audio frequency). But I don't work in audio and I don't have a good way to measure distortion.