nitramluap wrote:
It IS possible, actually and its omission makes no sense.
Yes, it is possible but it isn't desirable. Here's why:
For any linear digital audio encoding process, the bit depth determines the resolution of the volume levels the format can reproduce. For example, for 16 bit audio there are 65536 possible volume levels between dead silent & the absolute maximum possible level (digital clipping). For 24 bit audio there are about 16.7 million levels.
So let's say that you want to turn down the volume at the digital output. The only way to do this is by throwing away all of those levels louder than the desired volume by scaling the output to what is effectively a lower bit depth. One half as loud means the most significant bit ("MSB") will never be a one; half that loud & the next MSB won't be used, & so on.
At first, this doesn't seem so bad but the problem is the human ear doesn't interpret "one half as loud" as you might think: hearing is logarithmic (which is why we usually use the logarithmic unit "dB" to measure it), so a sound that is 1/2 the intensity of maximum is only slightly quieter to the ear. Depending on the maximum level the digital input device can produce, it could easily be necessary to scale the digital output to one ten thousandth or less of the maximum output just to achieve a comfortable level, & to scale it much smaller still for background music levels.
So, for an effective volume control with a usable range from "very quiet" to "full tilt" you would really be going from "miserably inadequate resolution" to "high quality resolution."
If you want to hear just how miserable sounding the quiet settings would be, & you have any sound editing software that can change digital volume levels of samples, try this:
Select a short sample & reduce it by 40 dB. Save it & then renormalize that saved sample to full scale. Listen to the result & compare it to the original. Not too pleasant, is it?