back
12 comments
I think MoFi deserves to be held to task for this, and I don't envy anyone who got misled about the product they paid their hard-earned money for.

Like the author, though, I'm still going to use this opportunity to try to confirm some information for anyone who still doubts it.

For listening purposes there is no discernable difference in accurate reproduction between 44100hz/24-bit PCM audio and any analog format. Indeed, there isn't an analog format that exists that doesn't add 1) Distortion imparted by the storage format and mechanical components, 2) a noise floor, from the same. So, one might even argue that analog is objectively worse if we are only talking about accurate reproduction. But let us say that we ignore those aspects and just consider the original frequencies in the master recording.

A standard CD can reproduce frequencies up to 22050hz, ample headroom for even a young child to hear everything. When played back, these get reconstructed to pure sine waves by the AD/DA converter and speakers, they are not "jagged" but pure tones all up and down. And 24 bits of depth is enough for 144dB of dynamic range - again, this is more than enough headroom to crank a track as loud as you could possibly bear without noticeably impacting the noise floor of your playback system.

Of course it's possible to go beyond these levels, but this is already more than enough to account for placebo.

Also worth noting that these days a vast majority of music producers and mixing engineers use digital recording. In fact, many of them now work almost purely in the box (that is, inside of software for arrangement, processing and mixing). This is because of convenience, flexibility, and power, but it's also because they recognize that digital audio has many concrete advantages in terms of accurate reproduction and archiving.

> these get reconstructed to pure sine waves by the AD/DA converter

That's not at all how DACs work. They create a stepped voltage output in some way, which is then filtered to remove the steppiness. Yes, if you're decoding a sequence of values that represent a sine wave, you'll probably end up with something functionally indistinguishable from a sine wave at the end, but 'reconstructed to pure sine waves' is nonsense.

Agree with everything else though.

It takes a familiarity with the Fourier transform to appreciate what jrajav is claiming.

The digitization introduces a high-frequency fundamental, harmonics of the fundamental, and additional artifacts that are at the sums and differences of these sampling frequencies and those in the signal. So long as the input is filtered so that it contains nothing above the half the sampling frequency (the Nyquist frequency), the artifacts introduced by digitization are all above the filter's cutoff frequency and can be removed from the reconstructed signal. At that point, if the original signal was a pure sine wave, then so is the reconstruction.

The technology is not perfect, of course - the digitization and its reversal is not arbitrarily accurate and the pure step filter does not exist - but, in practice, with a suitably high sampling rate, the infidelities can be reduced to the point where they are insignificant in comparison to those introduced by other parts of the process.

It is pointless to talk here about the digitization separately from the filtering, which is an essential part of the process, and it is pointless to argue that the filtered signal is not the original when the difference is undetectable by anyone. And if we were to be pedantic, then every signal is identical to a collection of one or more pure sine waves (possibly phase-shifted). Jrajav's statement is not nonsense.

Indeed, the nature of the fourier transform is what allows for the signal to be perfectly reconstructed in theory for frequencies below Nyquist.

More practically, it's not only filtering but also the slewing nature of both the analog part of a DAC as well as the physical speakers themselves.

I only brought this up specifically because I've seen more than a few times in "Analog is better" articles the 'stair step' representation of a jagged digital waveform, which is itself total nonsense.

I'm very familiar with the basic details of the Fourier Transform, Nyquist Theorem, etc etc. I was taught how to build a DAC from discrete components in University, and I've read up on the multiple types of DAC. And I still disagree over this minor point of wording - I still don't think it makes any sense.

But honestly, it really doesn't matter. You clearly know what you're talking about, wording or not, and I understand your argument and think it's sound, so no further nitpicking from me :)

You are, of course, entitled to your own opinion - even though you apparently know all the relevant facts!
The author offers a plausible explanation for early digital recordings gaining a bad reputation: the release of CDs from masters made for vinyl, without reversing the high-end enhancement those masters were made with in order to compensate for the losses that occur later in the process. As the author says, "the original master tape is not how those analog albums were meant to sound. The record is."

Making a vinyl recording from a master intended for high-fidelity (as opposed to lossily-compressed) digital reproduction will have the opposite problem.

All of this is an argument for matching the master to the medium, rather than for a one-step analogue process (to be clear, none of this justifies MoFi's deception - in fact, one can argue that it was not only being deceptive, but also deliberately exploiting some misunderstandings of its customers.)

Perhaps the company can settle the fraudulent advertising concerns by compensating customers with one of the high end "Audiophile rocks" that some people like. You know, the "Natural Crystal Pyramid" things you put on the amplifier or speaker to improve the sound. "$1,500 value!"
It's a very good point, the room and speakers, and even the variation in your ears through the day and day to day, probably have far more impact on what you hear than the recording.

OTOH perhaps we need a new format, where you can mix the audio locally, rather than just getting whatever the audio engineer decides.

Sound is inherently air in motion.

- The humidity in air changes its density and therefore the transmission of wiggly energy. Temperature probably does too, and there's always Doppler effects from running across the house.

- I tried to have phone calls in an empty house and people couldn't understand me, now almost all the walls are floor to ceiling with books and people don't understand me either over Zoom. You push wiggly energy, some of it comes back (or it destroys the building's structural foundations? I can't talk that loud).

- And there's also the whole issue of resonant frequencies. I know the resonant frequency of my head and can use it as reference to sing in tune if unaccompanied; I'm also able to entertain my child by finding the resonant frequency of corridors, elevators and so on. You can learn it too, just try to hum until one note is much louder and maybe rattles some bolted-on things. OF COURSE THE RESONANT FREQUENCY OF AN ENCLOSED ENVIRONMENT CHANGES HOW MUSIC SOUNDS, jesus. It changes how the same song sounds in different keys.

- And there's the issue of having absolute pitch, which, my dad tells me, ruins listening to music in the kitchen.

> OTOH perhaps we need a new format, where you can mix the audio locally, rather than just getting whatever the audio engineer decides.

FWIW, nothing stops you from appending additional processing to “remaster” for a specific playback/listening scenario — all radio stations do this, for example.

There’ve been a few attempts to deliver music in formats that would support remixing (the latest being Ye’s STEMPLAYER), but so far they’ve all failed.

... and the class action lawsuit is already underway - at least judging from what I see on my Facebook feed.