• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Audio foolery 2.0 - The rise and fall of objectivism

OP
nimar

nimar

Active Member
Forum Donor
Joined
Jan 25, 2021
Messages
213
Likes
216
Location
Ontario, Canada
I'm a believer in "44.1/16 is all you need for listening" but I like to see the current sample rate on my devices, because it lets me know if I have a misconfiguration.

(Nearly) all of my music is 44.1khz but occasionally I'll find that things are incorrectly set to 48khz or 96khz or something. So, I think it's a nice convenience feature if the DAC tells me what it's receiving.

Yeah, that was the category of "vaguely interesting" I had in mind.
 

Maki

Active Member
Joined
Oct 24, 2018
Messages
252
Likes
478
yes, they should be watching their gigantic VU meters
I've always found traditional VU meters horrendously ugly. Magic eye VU meters though, that's a different story.
 

Tokyo_John

Active Member
Forum Donor
Joined
Mar 6, 2021
Messages
214
Likes
289
For sure, you want a bit of headroom as we can safely assume no one is listening at max output. (which begs the question as to why measurements are taken at that level, a standard needs to be set but it could be 1/2 of max which is somewhat more likely, but that's another discussion). Which is why I've suggested all you need is to be comfortably above a certain threshold. Beyond that threshold you are back to chasing pointless numbers.

Agree with you, but the question is how much is “good enough?”

I'm sure someone else can give a more precise answer, but lets take the two DAC's we've been talking about.

D90 - 21 bits of dynamic range (126db)
ADI-2 FS - ~20 bits of dynamic range (120db)

I looked in my library and the absolutely highest DR music I could find is Carlos Kleiber's recording of Beethoven's 5/7th Symphonies at 21db DR. The vast majority of modern music is more like 2-5db DR. With Jazz somewhere around 3-10db and Concert (Classical) music 10-20db.

So in the absolutely most demanding case you have the ability to reduce the volume by ~100db and still maintain the full dynamic range of the music with the ADI-2 FS, ~106db with the D90.

Someone smarter than me can give precise values at normal 60-80 db listening levels but I'd say we are more than comfortably above the point where this matters in real listening.

It is important not to confuse THD+N requirements with the dynamic range of the music. Harmonic distortion and noise are always present, and the key is to render those inaudibly quiet relative to the lowest sound levels of the music you’re listening to. Some people seem to be more sensitive to harmonic distortion than others, but many say that 70-80 dB below listening levels is what the human ear can pick up. Let’s assume -80 dB is a comfortable margin. So in this case you want a system having THD+N that is -80 dB plus the -DR of the music. So if the music you’re listening to has 20 dB DR then you’ll want to aim for at least -100 dB THD+N. And I think the century mark is about the point where most people would agree that a DAC essentially eliminates distortion and noise.

Now if you are using digital attenuation on top of this, then you’ll want to add the amount of attenuation you’re typically using to the century mark in order to keep a comfortable margin and ensure that aliasing and other digital artifacts don’t creep in. Of course digital attenuators use dithering with noise-shaping and other techniques to keep the effects of attenuation smaller than the applied amount of attenuation, but again we are looking at getting a comfortably sure margin, beyond which point there are no practical gains to be made. So now we need a DAC that has a performance of -80 dB -DR of the music -digital attenuation.

In my system -25dB is a typical level of digital attenuation for ambient listening, which means that I want a DAC that gives me -125 dB performance in order to ensure that I have sufficient headroom and I don’t have to worry about distortion or noise from my DAC becoming audible. As you can see, the Topping D90 is just good enough to satisfy that threshold.

Now, if I were using an older DAC that had -70 dB performance I would almost certainly run a pre-amp and I wouldn’t rely upon the DAC for digital attenuation. And sensitive listeners would probably detect some distortion. So you see that the new capabilities of DACs today have expanded the range of way in which they can be used, we can push them further, and so we do. That’s human nature. So the question is whether there is a performance level beyond which we would never have any practical benefit, but that would be trying to predict the limits of our imagination and it isn’t a safe thing to do.

Still, the performance that is needed by a DAC or other device ought to be judged by its full potential application, and this can be done in specific instances as the need or possibility arises.
 

Tokyo_John

Active Member
Forum Donor
Joined
Mar 6, 2021
Messages
214
Likes
289
I'm a believer in "44.1/16 is all you need for listening" but I like to see the current sample rate on my devices, because it lets me know if I have a misconfiguration.

(Nearly) all of my music is 44.1khz but occasionally I'll find that things are incorrectly set to 48khz or 96khz or something. So, I think it's a nice convenience feature if the DAC tells me what it's receiving.

I think 44.1/16 should be good enough, in an ideal world. But the world is not ideal, and there are still significant differences among the strategies employed by the highest end DACs, how 44.1/16 files are rendered from the masters, etc.. 44.1 kHz is playing right at the edge of audible frequencies, and so there isn’t much headroom to allow for slop, and so it really raises the bar for everything to be done in the most optimal way, from pressing the digital copy to rendering it back to analogue in a DAC. The higher the rate, the more tolerance (headroom slop) the system has available to it, and the less exacting everything must be in order to produce a high quality result.

Note that “noise-shaping” and other processing approaches that take advantage of higher rates don’t require a sampling rate above 44.1 kHz in the original audio file. Even if you feed your DAC a 44.1 kHz signal, most of them scale up to a higher bit rate internally (you can also do this in software and players, to pre-match the DAC’s up-sampling frequency)...not for the purpose of reproducing audio at the associated frequencies, but rather to use it as a spectral work space where we can dump dithering noise and other artifacts at ultrasonic frequencies.

I’ve been trying to understand why somebody might pay thousands for the Chord M scaler, a computer that works really hard to extend an audio file way beyond audible frequencies. That extended signal is purely an invention of the software, it isn’t real. But somehow it is supposed to condition the digital audio in a way that helps the DAC get things right at 20 Hz and below. It seems difficult to believe that this is really necessary, it seems like a brute force approach to a problem that should have more mathematically elegant and less computationally intensive solutions.
 
OP
nimar

nimar

Active Member
Forum Donor
Joined
Jan 25, 2021
Messages
213
Likes
216
Location
Ontario, Canada
trying to understand why somebody might pay thousands for the Chord M scale

Don't have strong feelings about Chord one way or another, but just because people spend a lot of money on a HiFi product, that doesn't mean it does anything. See audio cables, power cables, power conditioning, magic rocks, etc etc etc.
 

mdunjic

Active Member
Joined
Dec 26, 2021
Messages
173
Likes
168
I've been interested in sound reproduction since I was a kid, and started to take audio a little more seriously about seven years ago when I bought my first "proper" stereo system. This comprised of an Audiolab M-DAC, a pair of Adam A7x monitors and some mid level AKG headphones (forget the model), all were purchased after extensive research on the many websites of the time. Around the same time I started buying CD's and did a clean sweep of any MP3/AAC music I had. Since then I've been content, after a few years I stopped listening to the headphones for one reason or another but the monitors and the DAC performed flawlessly.

Close the end of last year we had a child and my wife started eyeing my hard black angular Adam's with a certain look, after some time she informed me that they would be relegated to a dark corner of our home away from public gaze. The thing that was needed were some more home friendly speakers that had grills to keep prying fingers away from them. Thus the rekindling of the sickness began, the search for new HiFi equipment. As part of this search I stumbled across ASR and what began as a quest for some child friendly speakers turned into a full blown case of audiophile nervosa which demanded the full treatment, a completely new system.

Since my last bout of this illness there were now a lot more numbers. Everything was being measured. I read Toole’s book, I joined ASR and I started digging through the many threads to find the perfect setup. The DAC, amp and speakers to complete the system that would be the one true system, both child friendly and bring me closer to audio nirvana.

The Amp was pretty straight forward, a DIY Purifi Eval-1 kit. There was very little debate on this, both the objectivist and subjectivist were on the same page. Transparent amplification, ran cold and produced significant output. Measured well, sounded great. It has no buttons, no interface but its a power amp, it doesn’t need any of that.

Speakers, Revels. No questions asked. After reading Toole’s book I was already predisposed to liking them. I found a pair for a great price and couldn’t ask for more. Measured well, sounded great, again it is a passive component, it does what it is supposed to.

Before completing my perfect system I also purchased some Dan Clark headphones, measured great, sound amazing, do exactly what you expect.

DAC. The chart topping company with several products in the top ten was, you guessed it, Topping. Never heard of them before, and I was skeptical at first. But after reading through countless posts on a several threads on their top end DACs I won myself over. Surely, all these people praising these products couldn’t be wrong. They even had panthers on them playing golf, that’s got to count for something. I decided to go all out and get the best of the best, the Topping D90 MQA, using the latest and greatest chip on the market. The second best DAC ever reviewed according to the numbers at the time and still in the top five compared to the latest reviews. There were also some from Gustard and SMSL which looked promising but Topping had the longest track record of producing well regarded products so it seemed like the safe bet. In addition to the numbers on ASR there were plenty of more objectivist reviews using lots of random words praising this DAC.

As there were a few weeks lag between ordering and actually receiving the DAC in the interim period I ordered a second Topping DAC, the D70s as I decided that in addition to having a new family friendly system I would also replace my ageing M-DAC. Considering the design is almost nine years old and its limited to 24/96khz over USB it just had to be inferior to a new topping DAC and thus based on the numbers had to go.

The D90 arrived and what can I say, it sounded great. It produced beautiful music. Around this time I decided if I was going to replace my M-DAC, which has a built in headphone amp I also needed a dedicated headphone amp to use my lovely Dan Clark’s with, so I ordered an A90.

The A90 / a dedicated headphone amp was never a product class I had a need for before. It was only purchased to fill a gap in the D90 and allow for headphone listening. On paper it is in the top five headphone amps measured and can output deafening levels with almost any headphone.

Back to the D90. It sounds great but that’s where the pro’s end. Yes, the DAC sounds great, but it is inferior in every other way to my M-DAC which was designed nine years ago and honestly I couldn’t even say that’s a real pro as I’d be hard pressed to tell them apart. Its not a passive component, its something you interact with and it was bad at this.

M-DAC
Pros
  • Produces beautiful music over speakers / headphones
  • Program default input to start on or start on last used
  • Has a display with a volume meter so you actually know if a signal is being sent to the DAC and you forget to turn the speakers/amps on.
  • Automatically switches volume when plugging in headphones to last used level and back again
  • Has zero pop/click when turning on and off, plugged into powered monitors / AMP.
  • Volume control works over USB, full two way volume when used with Roon / Computer
  • Has a great remote with buttons that all do something useful.
  • Bit perfect test
  • Balance correction
  • Has a decent headphone amplifier.
  • Lots of other features.
Cons
  • Limited to 24/96khz over USB
  • Limited to 24/192khz over SPDIF
  • No DSD.
The D70s arrived and it was the exact same story, sounds great but that’s it. The A90 arrived and again, sounds great but its rudimentary to use and only ever turned volume to a quarter of full. Couldn’t think of a reason I’d need to deafen myself. Both products really only useful if used at your desk.

D90/D70s
PROS
  • Produces beautiful music
  • Can play 32/768khz PCM and DSD512 over USB
CONS
  • Poor UX from top to bottom
  • Remote is silly, several buttons do nothing, have to be very particular when pointing it
  • Pops when turning on / off
  • Hardly any programmable features, can’t set default input, can’t rename inputs
  • Doesn’t do volume control over USB, not even one way.
  • No headphone out, requires separate unit

A90
PROS
  • Produces beautiful music
  • Gets stupidly loud
CONS
  • Doesn’t adjust volume when switching between speakers / headphones
  • Draws 2.5W of electricity even when mechanical switch on the front is set to off?

Both the Topping and my old M-DAC produce beautiful music, but the Topping can play back some seriously hi-res files so I must be able to get more out of it. Thats it though, in every other regard the Topping products which were amazing on paper, in real life felt inferior to a nine year old product. Nine years is a long time in the current technology landscape.

It wasn’t long before I spent more time on ASR and worked out that DSD was a waste of time, so that’s one plus taken away from the Topping. And a little while later, reading more about sampling theorem that I came to the bitter realisation that 16/44khz is genuinely enough for music playback, if you really want to stretch it, then heck, make it 24/96khz. Funny, that’s what my old M-DAC was capable of doing over USB, that’s another plus taken away from the Topping. Leaving it with a grand total of zero advantages and a whole bunch of negatives.

What happened? How is it possible that the highest regarded products on ASR could be so disappointing. Let me be absolutely clear, they don’t sound disappointing, they sound great, but so does my old DAC.

The madness of subjectivism, where audio hardware was spoken about like a fine wine, has simply been replaced by the new madness of objectivism and the pursuit of “better” numbers.

Either the ability to play back files that have no good reason to exist, or to produce a SNR that is 10 or 20 db better than its predecessor, which is already beyond the ability of human hearing. Or to be able to power headphones to such a level that they could be used as speakers. Why? Why are we chasing these pointless numbers.

And I know some of you are shouting at your screens, “its to show good engineering stupid”, I know that. I appreciate the value in showing that the $10,000 DAC is objectively worse than the $1000 DAC. That’s grand, but things are going too far, there’s a point where its just a game, and most likely it is a game that is being hacked. What design choices are being made by Topping et al just to get good numbers, without considering long term stability, energy consumption, safe design etc etc.

I appreciate that Amir states in his reviews that he is only recommending products based on the numbers and can’t speak for long term reliability etc. Though I don’t think its enough, somewhere on the front page of ASR should be a “Read this first” explaining that chasing numbers if entirely pointless beyond certain thresholds. To be fair there are several posts covering all these topics, they just take a bit of digging to find.

So where did I end up. I’ve sold all my Topping gear, returned to using my M-DAC and bought an RME ADI-FS 2. It has slightly worse SNR to the Topping products but has all the same great functionality of the M-DAC plus a whole lot more. The crazy thing is that it was even cheaper than the Topping stack.
Beautiful post! Enjoyed reading it … even as hard core objectivist, I continue to use my Quad CDP-2 as dedicated DAC to this very day … it sounds amazing, and I really don’t care that much about higher res than BASIC CD quality WAV/AIFF/FLAC tbh, as 99.99% of my own ripped music is such.
 

mdunjic

Active Member
Joined
Dec 26, 2021
Messages
173
Likes
168
Being close to 60s myself, and after spending almost 40yrs playing with hifi, I have learned that every room has a limit of SPL it can take, based on its shape, acoustic treatment and speaker placement/positioning

Electronics definitely matter (as long as they have good measurements i.e. flat frequency response, wide enough bandwidth, distortion levels consistent below at least 90db, low jitter, etc) but not as much, imo, as room and speaker interaction, which is the key

I was able to get to the basically almost the same (or very similar) sound signature in my same room, with both similarly measuring Rotel and Quad electronics … as long as it is driving exactly the same set of speakers, positioned the same in the same room, at similar spl at 1m away from speakers … of course since these amps are similar BUT still don’t measure exactly the same, I developed preference for Quad sound over time and that’s where I was in last 15 yrs (still think many Rotels are more than decent and more than acceptable electronics for 90% of hobbyists) … what almost always mattered more, in my experience, than slight difference in electronics, was room acoustics and proper speaker placement
 
Last edited:

xnor

Active Member
Joined
Jan 12, 2022
Messages
193
Likes
207
Thank you for posting that @ op. People don't realize how bad human hearing is and how easy it is to get extremely well performing devices for cheap these days. The push for lower numbers is simply one way to sustain the legitimacy of the hobby.

But part of being objective about audio is realizing exactly that. Psychoacoustics should be as important as raw device measurements.

Regarding the pros and cons you found: several of those things can and should be tested for and included in reviews. Most (all?) of them are objectively measurable:
- Programmable default input
- Volume meter
- Separate gain for speaker/headphone out
- Pops/clicks (who doesn't remember the videos of Schiit amps damaging headphone drivers?)
- Balance control
- Idle power draw
- Efficiency
...


Shopping just by a few audio performance numbers is never a good idea when you expect more than just that raw performance, otherwise you can end up with something completely impractical for your use cases.

To anyone new to this, just treat raw audio performance measurements as a tool to figure out whether or not a device has serious design issues. You can start hunting for lower numbers once all your other criteria are met.
 

mdunjic

Active Member
Joined
Dec 26, 2021
Messages
173
Likes
168
Benchmark LA4 beautifully driving 2 Quad 909s and Neat Motive 1s … Quad CDP-2 as DAC and Lenovo laptop as Roon Core ✅
 

Attachments

  • 2FA3708C-47C9-4EDE-A47F-08FF6D4C992D.jpeg
    2FA3708C-47C9-4EDE-A47F-08FF6D4C992D.jpeg
    317.7 KB · Views: 76
  • 711FF86A-E8E3-418C-955A-575EA4C1C063.jpeg
    711FF86A-E8E3-418C-955A-575EA4C1C063.jpeg
    352.8 KB · Views: 81
  • AE3994E1-8E4C-454C-AE49-28BBF204B205.jpeg
    AE3994E1-8E4C-454C-AE49-28BBF204B205.jpeg
    271.1 KB · Views: 75
Top Bottom