The current fashion in digital audio right now is "bit depth". As has been the standard practice for decades, audio is being sold by the numbers, whether meaningful or meaningless. And of course, everybody involved in digital audio transfer is now shouting about their bit depth. 16-bit? 24-bit? 32-bit? Hey, why not 64-bit while they're at it? Logic would suggest that the greater the bit depth, the better the sound. Accordingly, LP to CD transfer companies always point out their 24-bit process to make you think their transfers will sound better. But is there a genuine advantage to higher bit-depth transfers?
To answer that question, let's look at how bit depth relates to digital audio. The function of bit depth is to determine dynamic range. A greater bit-depth gives you more potential numbers between the zero-crossing point of the waveform and the peak, thus greater amplitude is possible. Or, to put it another way, a greater differential between the peak level and the noise floor. 24-bit might be, theoretically, quieter than 16 bit (which is already dead quiet). The popular way of thinking is that 24-bit has "higher resolution" than 16-bit, but this is fallacy. Resolution is determined by sampling frequency, not bit depth. To illustrate, picture 44,100 orange crates standing in a row. Those crates represent one second of CD audio. Bit depth measures the size of those orange crates. A 16-bit crate can hold 65,535 oranges, and a 24-bit crate can hold 16,777,216 oranges. Thing is, even if they held a TRILLION oranges, there's only ever going to be 44,100 of them. The resolution remains unchanged.
The question then becomes, just how many bits do we need? The dynamic range of 16-bit digital recordings is around 90 dB. Dead quiet to full-blown symphony. But we must consider how much of that 90 dB dynamic range we're actually going to use when transferring albums to CD. The best turntables are going to be in the 70's. The very best vinyl is going to be in the 50's if you're lucky, with most commercial releases hovering in the 40s. Yep. 40's. So - we only need a about half of the 90 dB dynamic range that 16-bit makes available to us. Why don't LP-to-CD services mention this little detail while touting their 24-bit transfers?
So now let's look at the resolution issue. The standard for CD audio is 44,100 kHz. This will theoretically give response to 22,050 Hz, according to the Nyquist limit. It is extended to 22,050 Hz to yield a flat composite response to 20,000 Hz when the multi-pole low-pass filter is factored in. Otherwise you would hear the quantization noise. But how do we listen to music? We cannot wire sound directly into our brains, we must use speakers. A speaker cone cannot stop at one point, then magically re-appear at another. It has no choice but to travel through all intermediary points as it goes from peak to peak. Thus, a speaker creates it's own infinite level of quantization while it reproduces the source. They don't mention that either, do they?
But here's the most puzzling thing of all. The common blurb for transfer websites is "Our transfers have higher resolution because we record at 24 bits at such-and-such kHz and then downsample to make the CD." Great, but for the fact that whenever you resample, you ADD NOISE. So what they're saying is "Our transfers sound better because we add noise to the signal."