The audibility of defects is
complicated and I'm not an expert. You need graphs & curves to define it... You can't nail it down to a simple number, although we can usually say that -120dB SINAD is good-enough. CDs have a dynamic range of 96dB but that's generally better than human hearing even if we are listening at 120dB "rock concert levels".
Negative feedback or not,
no amplifier (or other electronics) should have audible distortion unless it's over-driven into clipping.
The quietest sound we can hear is about 0dB SPL so a signal-to-noise ratio of 120dB means your music can hit 120dB SPL and when it drops to silence you won't hear any noise.* But in reality the acoustic background noise in your room is probably around 30dB so you'll never hear the background noise (near 0dB SPL) from the speakers. And the nature of the noise matters too... If you have a 1kHz tone at 25dB and acoustic white or pink noise at 30dB, the 1kHz tone won't be completely drowned-out and you'll hear the tone.
Your hearing is also not "flat". It's most sensitive at around 2kHz so you're more likely to hear noise at mid-frequencies.
Your hearing is also most-sensitive to harmonic distortion in the mid-frequencies. And when you get to the highest frequencies you can't hear the harmonics at all! If you can hear to 20kHz and you have an 11kHz signal with lots of harmonic distortion, the 1st harmonic is 22kHz and you can't hear it. You won't be able to tell a square wave from a sine wave! Music (or any natural sound) contains lots of harmonics and that means that you usually can't hear moderate amounts of harmonic distortion even if you can hear the distortion with pure test-tones.
There are other kinds of non-harmonic distortion and those can sound worse, but harmonic distortion (from clipping) is the most common kind of distortion. Clipping generates harmonic distortion, and only odd harmonics. That means if you can hear to 20kHz, you can't hear clipping of any frequencies above 6.6kHz.
And it's true that the higher-order harmonics can be worse. If you have a 100Hz tone with a harmonic at 200Hz it's more likely to be drowned-out that a 1kHz harmonic at the same level. ...The more spread-apart two (or more) simultaneous tones are, the more likely you are to hear them as separate tones.
(Ergo negative feedback is inherently bad.)
Anyone who's studied engineering knows
that's just wrong! There's a LOT of nonsense in the "audiophile" community!
Negative feedback ("corrective feedback") makes EVERYTHING better. (At least it CAN when done properly.) It can give you lower distortion, lower noise, and flatter frequency response.
Positive feedback is BAD. It's what causes that feedback squeal in a PA system when the microphone picks-up the sound from the speakers and re-amplifies it in an infinite loop. (Positive feedback can be good when you're building an oscillator.)
Negative feedback is how "everything works"... Negative feedback is the steering corrections you make while driving... You can't just point the car straight and expect it to go straight down the road. It's how cruise control holds a steady speed and it's how our self-driving cars will work in the future. Feedback is how airplanes find the airport and how ships find the harbor. It's how your home thermostat works, turning the heat on when it's cold and turning it off when it reaches the target temperature. (My car doesn't have a thermostat. Some newer cars do, but in my car I'm the thermostat and it's up to me to turn the heat up or down or turn the heat off and turn on the AC when it's hot.)
* Plus, you'll probably have a temporary threshold shift (temporary deafness) so you are less able to hear quiet background noise after exposure to loud sound.