Imagine measuring someone's height, but your ruler only has marks every inch. Someone who's actually 5'7.3" gets rounded to 5'7". That little sliver between the real height and the rounded number is the error you introduced just by having a ruler with limited resolution. Digital audio does the exact same thing to sound: a PCM recording measures the wave's amplitude thousands of times per second, but it can only store each measurement as one of a finite number of values (set by the bit depth). Whatever doesn't fit gets rounded, and that rounding error is quantization noise.
Why bit depth is the dial that controls it
More bits means more possible values to round to, which means a finer ruler and less rounding error per sample. A 16-bit recording (CD quality) has 65,536 possible values per sample; a 24-bit recording (common for production masters) has over 16 million. Both will have some quantization noise, since rounding never fully disappears, but the 24-bit version's error is so much smaller it sits far below what a human ear can perceive.
Here's the genuinely surprising part: quantization noise sounds like a faint, grainy hiss, not like distortion, and it shows up most audibly during very quiet passages (a fade-out, a held synth pad) where there's nothing louder to mask it. It's effectively a tiny, mathematically inevitable tax that low-level detail has to pay.
Why it matters for DJs
Quantization noise is one more reason a 16-bit file behaves slightly differently from a 24-bit file on a big system, even before lossy compression enters the picture. It's also the reason quiet outros, ambient breaks, or anything with a low noise floor can sound subtly grainier on lower-bit-depth files once you're cranking volume on a system with real headroom.
In the context of Headroom
Quantization noise comes up in Module 1: Sound right alongside PCM and perceptual coding, rounding out the picture of what actually happens to a sound wave on its way into a digital file.