Originally Posted by JohnFerrier
But I can't hear amplifier noise without power conditioning. What good does reducing it by 30%, 100%, or 500% do? If you hear noise without power conditioning, your system is not well designed. (Well designed doesn't mean expensive either.)
What you stated is a misconception.
Yes, the noise in your audio systems should be below your listening threshold. In other words, you should not hear speaker hiss at your listening position or hiss in your headphone. But once you play some signal through the system, the noise is added to the signal you hear. Say your signal is a smooth sine wave of 100Hz, and if you add a bit of 5000 Hz noise to it you start to see jagged curves. Even if your ear can not hear the noise by itself, your ear can hear the difference between a smooth 100 Hz sine wave with a jagged one.
Here is a real example. When you make a 24 bit recording and map it down to 16 bit, it is extremely apply dithering (noise shaping) to the last (16th) bit instead of doing simple truncation. Famous dithering algorithms include SBM (Sony), UV22 (Apogee) and K2 (JVC). Can you hear the single-bit signal (0000,0000,0000,0001=-96 dBFS) playing in your CD system? No, unless you turn volume way up. However, when this last bit is a part of the music averaging at -20 dBFS or so, it certainly can make a difference. If noise dithering is not applied to the last bit in CD, the result is harshness, and the loss of detail and spatial information. In fact, with proper noise dithering, you can extend the dynamic range of CD into 18-19 bit (a well known fact to mastering engineers). It is because humans can hear signals below the noise floor, a fact appreciated by LP lovers everyday. Despite the poor measured S/N ratio of LPs, the useful dynamic range is much larger than what S/N suggests, because we can hear signals buried within LP's analog hiss (especially because the noise in LP playback is non-random). In the early days of CD sales, dithering was not as advanced as it is today, and CD sound quality suffered.
You can't hear the digital 1-bit truncation noise being played alone, but it degrades the music you can hear. By the same virtue, even if you can't hear the analog noise of your amplifier without input, it can still audibly degrade your music when there is input.
Do I need to offer a simpler analogy? There are two pieces of white paper, but only one has some faint stain marks. When the room is very dark, you can't see any stain mark and you declare both papers are blank. With better lighting you start to see the difference, and you know which paper is cleaner.