To be extremely pedantic (and annoying), the bit-depth doesn't strictly determine the dynamic range - it determines the number of discrete steps of loudness that can be reproduced. 8-bit samples give you 256 steps of loudness. 16-bit gives you 65,536 steps. 24-bit gives you 16,777,216 steps.For your information, here are some facts for those interestred:
- In the audio world "lossless" typically refers to lossless relative to CD quality, which is 16/44.1 or a dynamic range of 16 bits and a sample rate of 44.1kHz.
- The sample rate determines the maximum frequency you can represent. A digital signal is a discrete (made up of samples) representation of a continues signal (waves). To reconstruct a sinus wave you need exactly two samples. This means that the maximum frequency you can reconstruct equals half the sample rate. This reconstruction is exact and not an approximation (as it is for image pixels). In other words, with a sample rate of 44.1 we can accuratly reconstruct frequencies up to 22kHz, well above the limit of human hearring. For reference, the highest note on a piano is 4286 Hz and most speakers will not be able to produce signals over 22kHz either.
- Does playing at 192kHz make sense? Yes, if you like to play music for your dog and you have very high-end speakers with no other bottlnecks in the connection chain. Otherwise, absolutely not.
- The dynamic range determines the number of different volume levels you can represent. With 16bit you can represent signals from wispering to over 90dB, enough to cause damage to your ears after long exposure.
- Is there an advantage of a dynamic range higher than 16bit? Yes, if you want to accuratly represent fine details ranging from whispering to explosions. For most pop/rock music there is no difference at all.
- Recordings are mostly done at higher sample rates and higher bit rates. Why? This is not because we can hear a difference in the recording, but because it gives additional headroom during production, changing a signal invitably results in some losses which can as such be minimized.
- Airplay does support ALAC 16/44.1, in fact if I am not mistaken, it transcodes all input to this format for transmission. I don't see any reason why HomePods would not be able to play lossless input streams. If you will hear a difference is another question...
- Some people seem to belief everything lossy is the same, this is obviously not the case, the codec and bitrate make a huge difference.
- Currently Apple uses 256 AAC, truth is, most people don't hear a difference with lossless (CD quality) either, especially with low end equipment like HomePods for example. However, there are definetly people who can hear a difference on high-end equipment. If you want to check for yourself with your equipment you can do an ABX test here: http://abx.digitalfeed.net/itunes.html
- There are many reasons why you can compress a PCM signal lossy without any perceptual difference at all. For example, our sensitivity does not only depends on the signal intensity but also on the frequency. For example, humans can not hear sounds at 60Hz under 40dB. While these signals are encoded in PCM, these can be removed without any perceptual difference for humans.
- Eventhough most people can't hear a difference in a scientific ABX test they still belief they do hear a difference. Why? One reason is because mostly they don't test blind. At the moment you have prior knowledge you can't do an unbiased test. Tests have been done with exactly the same equipment audio but different logo's (Bose vs B&O for example), the more premium brand will consistenly perceived better even if the hardware is exactly the same. Secondly, it is common to decode the signal sligtly different. For example, simply increase the volume with 1dB and almost all test subjects will perceive this as higher quality. Third, often tests are done where other factors or at play, such as the DAC, connections and so on.
What a higher bit-depth gives you is the ability to represent a higher dynamic range without being able to hear distinct steps in loudness.
There's a direct analogy between this and digital images (and viewing devices such as TVs, monitors, etc) - non-HDR content would typically be displayed at 8-bits per color channel (RGB), giving us 256 steps between zero and maximum brightness. HDR content (usually 10-bits per channel), would give you 1024 steps of brightness.
The advantage of having a higher bit depth (or more steps) is less visible banding when showing a gradient from extreme brightness to extreme darkness.
If you have a TV or monitor capable of recreating a scene with a high dynamic range of brightness as your eyes would see in real life - e.g. a shot of a campfire where you're able to look into the flames but also the black cat hiding behind someone in the shadows, the 10-bit depth would allow enough steps of brightness to display a smooth gradient between the light and dark parts of the flames, and the texture of the cat's fur. If you only had 8-bits, the campfire would be a big white ball and the cat would look like a very dark gray stencil (if you can see it at all).
Coming back to audio, if you had an 8-bit recording and still wanted to capture someone whispering next to a rocket, you could but the whisper would sound like a bad Skype call with sudden jumps in loudness (it would sound really scratchy as it cuts in and out of the lower end of our dynamic range), while the rocket would also have weird jumps between loud and super loud (and would also sound distorted if it clipped the higher end of our dynamic range).
A 16-bit audio clip would be able to capture this range, but you'd probably want to record the source at 24-bits so you have headroom to make stuff louder, softer or to compress the loudness in post production to fit the 16-bit output file.
I'll keep quiet now.