I’ve read right through this thread and, as usual, the main arguments are about whether we can capture and reform the sampled audio waveform accurately with 16-bit and, therefore, do we really need 24-bit resolution?
I see exactly the same arguments around the 44KHz sampling frequency, namely is it worth capturing frequencies above 20kHz (the usual ‘only bats, dogs and small babies can hear higher’ is often trotted out).
All these arguments, to me, miss the point which is that, once we turn from an analogue method of recording, storage and reproduction, all the rules about signal to noise and frequency response become fairly irrelevant unless, of course, the system truncates what we need.
With digital what matters most is not how accurately the waveform is stored, given that we’ve adopted the 16-bit/44kHz CD Red Book system as some sort of hi-fi standard, but what happens when we record, store and reproduce the waveform using a system that operates at this resolution.
By this I mean, what do the digital systems do when recording, storing and reproducing an analogue waveform that impinges on what we hear?
For example, the question regarding sampling frequency should not be about whether recording frequencies beyond 20kHz matters but what do analogue to digital and digital to analogue converters do at and beyond 22kHz (based on a 44kHz sampling limit)? The answer is that the distortion and noise produced beyond 22kHz is horrendous and must be heavily filtered if we aren’t going to scream in agony and drive our amplifiers crazy! This also draws into question what distortions the brick-wall filters at 22kHz introduce too? These are arguments not pertinent to this thread, but I just want you to think about them.
Because when we talk about bit depth we should be looking at the problems involved in digital recording and reproduction
Let’s start with the basics. When recording engineers were limited to analogue tape they would frequently push levels to +3dB simply to minimize noise at the expense of mild distortion. With digital, I believe, engineers no longer monitor recording levels at all closely in the belief that the digital system is noise free.
I’ve been in recording studios where recordings are made peaking at -20dB just to make sure that sudden unexpected transients don’t hit the 0dB level. Anything above 0dB is pure, nasty distortion and so must be avoided at all costs.
If we say that, for orchestral music, we are looking at 60dB dynamic range, then we need a digital system that remains clean for 80dB overall. Now this should be within the range of 16-bit systems but the problem is that the lowest digital levels a prone to quantization noise, which is why engineers quickly added dither so that we would never hear this rubbish.
Add to this the problem that digital systems are not as perfect as we think they are - for example most 16-bit systems actually have trouble working to better than 14-bit accuracy when you measure them - and there’s an advantage to using a greater bit depth.
On replay modern, high resolution DACs are often working at 32-bit mathematical capability in order to minimize distortion and noise and deliver superior accuracy. These DACs seem to be able to work slightly better when given a 24-bit file than 16-bit (and I’m ignoring DSD here, for the moment, which has different advantages and problems).
In essence the real problem is that we are using numerical coding methods to try and record, store and reproduce a complex analogue waveform, so the usual rules accorded to the analogue markers of signal-to-noise and frequency response don’t apply.
I’ve spent time working with digital engineers and they tell me that what goes on inside ADCs and DACs isn’t as clear cut and accurate as the layman would like to think. On that basis the higher the bit depth the more likely it is that we can capture, store and reproduce the musical waveform without noise and distortion intruding.
So is 24-bit ‘high resolution’ compared to 16-bit? My answer would be yes. However that was not the question. It was closer to ‘is a 24-bit/44.1kHz file truly high resolution’? That’s not so clear on the basis that the noise and distortions introduced by the 44kHz sampling may well outweight the advantages of the greater bit depth.
