> If you are a fan of music, especially guitar-based music like classic rock, there is no substitute for a vinyl record created through a 100% analog mastering process. Analog captures harmonics that digital simply can't. Analog mastering also has much wider dynamic range than does digital mastering.

Analog mastering doesn't have anywhere near the dynamic range of digital; a CD with its paltry 16 bit has a noise floor below -90dB. There are staggeringly few analog mastering setups that can do that. Move to 20+ bits (i.e. all pro digital mastering setups) and you are getting to the point where an analog setup might be able to maintain the SNR, but would be astronomically expensive.

As far as the harmonics, there is little evidence that humans can detect 22kHz+ sound at safe listening levels, and once you talk about 96 samples/s most speakers can't reproduce anything like the original signal at 40kHz+. It might make a physical difference if you have a $20k electrostatic setup, but you won't be able to hear it.

As for the skeptics here, NPR has a pretty good test so you can see for yourself whether you can hear compressed vs high-range audio:

http://www.npr.org/sections/therecord/2015/06/02/411473508/h...

> Analog mastering also has much wider dynamic range than does digital mastering.

This tends to be true in practice due to "loudness war" problems - many CDs are mastered not for accuracy but for popularity.

What's the typical line-level input voltage for an audio signal? Something like two volts? Four, peak-to-peak? A 16-bit ADC quantizes that four volts into 65,536 distinct quantities which are reproduced as voltage levels and hence, sound by a DAC on playback. Okay, fine, whatever. BUT... in a 100% analog process, the line-level input max is much greater than two volts. I've heard of analog equipment being able to register signals in excess of 70 V. This gives the audio engineer practically infinite headroom in which to record and mix sound without losing detail.

Where analog generally falls down on the job is, you typically need actual musicians playing actual instruments in an actual studio to lay down a track. None of this wave-table-synth, virtual instrument shit. And no Auto-Tune or other automatic correction either; if a musician flubs a track, generally they have to rerecord it. Pop producers consider this a bug; serious musicians consider it a feature.

And a 24 bit ADC that quantizes the lower 16 bits into 4Vpp would be capable of registering signals at 512Vpp. No analog devices have such a large headroom (granted they don't build such a device, but if you attenuate the signal at the entrance to the ADC, you get something similar).

This, btw. is why nobody masters at 16 bits; the insufficient headroom. 16 bits is plenty for playback though.

[edit]

I do want to acknowledge the big weakness of a digital setup that you touch on here, which is the hard clipping. If you ever clip in digital, you have ruined the recording while many analog setups will have a soft rolloff (Tube amps v. BJT amps are similar; the soft clipping of a tube amp is much less unpleasant than the hard clipping of a BJT amp. You can "get away" with a lower wattage tube amp for this reason; a properly sized BJT amp is more accurate though). All this is moot for professional recordings though as a sound engineer who regularly clips by accident (rather than to intentionally get a particular sound) will get fired.

> No analog devices have such a large headroom

Obviously you've never mastered a Disaster Area album. :)

Voltage isn't particularly relevant of itself; what's that in dB above the noise floor?

Precision is so boring though. It's like mapping the inside of a cathedral with lidar or something; yeah it's awesomely accurate but it's not substitute for just walking in and enjoying the spectacle. You won't have a better aesthetic experience from knowing the position of every individual block of stone with micrometer precision.

Not an equivalent comparison.

Perhaps a grainy photograph vs. the lidar is a more apt comparison. The former has character, but the latter is a much better approximation of the real thing. Also you can generate a grainy photograph from your accurate digital measurements if you so choose.

Why a grainy photograph and not a good quality one? I feel you're begging the question here by assuming your conclusion as your premise.

Maybe a better metaphor would be a laser pointed vs a flashlight - assuming they emit the name # of lumens, which one is more useful for exploring a dark room? This is why I think the pursuit of endless precision in calculating the value of irrational/transcendental numbers is an amusing hobby but ultimately a waste of time.

The frequency response of microphones generally plummets at (or below) 20 kilohertz.

Digital sampling can capture 20 kilohertz just fine. An analog recording pathway isn't really going to capture much more information than a digital one.

Not only that, filtering above the audible range is a feature. Ultrasound can give audible artifacts, for example intermodulation between close tones. And all sorts of high frequency noise exists even from the equipment itself. (This is just one reason 96 kHz material doesn't make sense for the consumer. Any good playback chain would filter it out.)