Klippel distortion tests (two-tone 70hz 800hz tests). Where do you fit, as compared to +8680 participants?

[Pdxwayne]

Average at best for me. Same result in two tries: -33 dB. Sony QC 35 II from an iPhone over Bluetooth.

Rick “maybe the subtler distortion products are masked by my tinnitus ” Denney

Thanks for checking!

I wonder if you would do better if you check using hardwire to a nice dac+amp combo.

 

[rdenney]

Thanks for checking!

I wonder if you would do better if you check using hardwire to a nice dac+amp combo.

Maybe. But I bet not. I have to be at home to do that, though, and I’m traveling this week.

The music examples are execrable—I had tried their test before but couldn’t tolerate the music long enough to get to a conclusion.

Rick “who’d be using K371’s in a JDS Atom amp at home” Denney

 

[Pdxwayne]

Maybe. But I bet not. I have to be at home to do that, though, and I’m traveling this week.

The music examples are execrable—I had tried their test before but couldn’t tolerate the music long enough to get to a conclusion.

Rick “who’d be using K371’s in a JDS Atom amp at home” Denney

Yeah, solderdude and I have the same opinion about the music examples.
: )

 

[Deleted member 48726]

Just reviving this thread as it was linked to in another one.

Laptop with Bluetooth connected Sony WH1000 XM-4 with noise cancellation on (kids watching TV) and low-volume as I hate tone tests.
Two-tone: -39 dB
-First try. Don't like tone tests so I won't do it again.

The very crappy Tracy Chapman test: -33 dB
-Tried two times. Honestly it sounded a bit distorted all the time so it gets hard fast.

 

[tmuikku]

Tried the two tone tests, smack dap average performance with both headphones and speakers. Sorry, don't want to listen the chapman tracks

It was interesting to test Bl(x), Le(x) and Cms(x). Bl was most audible, at least result got -45db or something, with Le -39db or something like that, bit less annoying sound. Cms(x) I got -15db, by far the least offensive sound of the three. I'm not sure though how the tests were actually balanced and how these are comparable with each other, the Cms test says 0db was "real speaker" while the other two didn't I think.

Anyway, practical takeaway for me is that about >1% speaker distortion is likely audible to me (average person) especially if it's from driver motor, Le(x) or Bl(x), which are 3rd order harmonic mostly. While < 1% distortion is likely not that meaningful, especially if it's mostly from suspension. Cold be simplified that when speaker distortion rises to 1% or more the sound starts to change for average person. Then, a target for nice hifi system performance could be <1% (~-40db 3rd harmonic) with party SPL levels.

 

[MacClintock]

I got -51 dB, on my Mac with an IEM and much luck. The funny thing is, the master of audiophiles, GoldenSound, recently on a video claimed to having reached -75dB. I am not saying that he is lying, but I am pretty sure he is!

 

[pma]

-60dB, with my DIY Visaton W100X + Monacor HT-88 small 2-way speaker driven from Topping D10s + my A250W4R amplifier.

 

[Mr. Haelscheir]

You folks have quite some ears. The best I got for the title test a few hours ago was -48 dB with an EQed Meze Elite (which I had already measured as having exceptional multi-tone distortion; looking at FFTs of the Klippel test signal, it seems like BI multi-tone distortion is the closest in character to the kind I've been measuring out of headphones). I had previously scored -42 dB (sidebands around 44 dB below the fundamental per my FFT analyses). Somewhat surprisingly, putting my desktop to sleep to silence the fans and instead connecting my cellphone to my FiiO K9 Pro ESS for unEQed listening had me scoring only -39 dB. I can hear the harmonics or subtle "warbles", but they eventually just disappear or get masked and I end up hearing the same artifacts or "placebo" for each. For curiosity, I used Equalizer APO to apply notch filters (whether or not any of the folks who made it to -69 dB used this trick or looked at the FFTs among other exploits) which allowed me to get to -63 dB, my needing to apply yet deeper peaking filters to be able to discern -69 dB (sidebands 71 dB below the fundamental, or around 0.03% THD, my FFT showing around an 81 dB SINAD for the undistorted signal; with the -100 dB 999 Q peaking filters, the upper tone's sidebands were instead 23 dB below the fundamental, almost at the -120 dBFS noise or distortion floor of my MOTU M2 loopback as seen on the REW RTA), the volume in both cases rendering the recording's noise floor audible.

For ""6 inch driver" playing "octave - multitone - Bl only"", I scored -21 dB (though I was technically guessing by -18 dB which per the FFT had the sidebands around 34 dB below the fundamental). On my REW RTA FFT through the MOTU M2 loopback with 128k length and Blackman-Harris 7 window, this recording consistently already had some sidebands around 58 dB below the fundamental, whereby the -36 dB distortion level was almost indistinguishable in the FFT, -42 dB looking properly indistinguishable, however folks could have gotten to that level or the -45 dB maximum without exploits.

Otherwise, for the ""full range speaker" playing "Music T. Chapman"" test, I swear the recording is so poor that by -15 dB, the added distortion is indistinguishable. I don't know if there were any tells or exploits that have since been fixed, as I'm hearing the same decays, artifacts, or subtle inflections in both by that level with all distortion buffeting having disappeared.

Now, I do unfortunately have high-frequency tinnitus and have been attending lots of classical symphonic concerts, but since my first encounters with sine sweeps, I would swear that even when I measure harmonic distortion well below what should be audible, I hear spurious tones being excited in addition to the actual perceivable fundamental suggesting either my own ears distorting or my cochlea having spurious excitations, perhaps further masking the harmonics I should be looking for. E.g. If I play an 800 Hz pure tone that certainly doesn't have digital distortions above the FFT noise floor, I also hear a 930 Hz resonance at least with the left ear, maybe a different frequency for the right ear, said resonance being proportional to the fundamental's loudness and for my moderate to loud listening starting to just become audible at as low as 170 Hz. But once I get up to 930 Hz, I start hearing a 1,080 Hz phantom resonance or something, and so on. I haven't read into whether and to what extent different individuals' ear physiologies may distort audio to different levels or whether some individuals actually have sharper spectral filtering.

Of course I am still able to have blissful musical experiences, but I can't help but wonder how life would be without this probably primarily fire-alarm-induced tinnitus and these weird single-tone resonances.

 

[HaveMeterWillTravel]

So I just learned about and tried it for the first time.
However, the test keeps stopping for me at -45dB even without any 'errors'?
I have Full Range Speaker, Chapman and headphones selected.

Any ideas?

 

[Mr. Haelscheir]

So I just learned about and tried it for the first time.
However, the test keeps stopping for me at -45dB even without any 'errors'?
I have Full Range Speaker, Chapman and headphones selected.
View attachment 418355

Any ideas?
View attachment 418355

Is it showing correct even if you select the option you know to be incorrect? Otherwise, different test sounds have different minimum thresholds before the test is ended.

 

[HaveMeterWillTravel]

Is it showing correct even if you select the option you know to be incorrect? Otherwise, different test sounds have different minimum thresholds before the test is ended.

That was a good thought. But no, when I intentionally choose the 'wrong' answer, it logs it appropriately.

 

[Mr. Haelscheir]

That was a good thought. But no, when I intentionally choose the 'wrong' answer, it logs it appropriately.
View attachment 418368

In that case, you have impressive hearing at least compared to how I do with that track and my tinnitus. Care to share what things you particularly listen for?

 

[MAB]

In that case, you have impressive hearing at least compared to how I do with that track and my tinnitus. Care to share what things you particularly listen for?

Klippel gives a good summary of the test result, including summary statistics on where your score lies within the population, and if you generated a score that is an outlier.

For instance, looks like lots of people can generate a score -69dB on the two-tone full-range speaker test. Lots of people are... outliers.

 

[HaveMeterWillTravel]

Klippel gives a good summary of the test result, including summary statistics on where your score lies within the population, and if you generated a score that is an outlier.

<SNIP>
For instance, looks like lots of people can generate a score -69dB on the two-tone full-range speaker test. Lots of people are... outliers.

Initially it gave me the fits. Why? I kept on selecting what I thought was the 'non-distorted' selection. Old habits die hard.

Your comment reminds me of Monty Python, "You're all different". Crowd in unison: "We're all different!!" and one lone voice speaks up "I'm not."
But instead, "You're all outliers".

 

[MAB]

View attachment 418472

In that case, you have impressive hearing at least compared to how I do with that track and my tinnitus. Care to share what things you particularly listen for?

Longer answer...
In each of these studies, you will see a large number of outliers. We expect about 10 people in this population to hear combined distortion at -69dB, but in fact we see 305 people who appear to have maxed out the test.

Either an unusually large population have outlier-great hearing, or they have good analysis tools.

It's easy to discern -69dB by looking at the spectra:

I can do no better than -51dB on this test, and it is a bit exhausting, boredom being part of it. Some of these help at higher volume, which is a bit self-limiting. Each person's ear will affect the result depending on the distortion added, Klippel offers just a few modes, each person will have their own combinations they are particularly sensitive to.

More interesting than ego-stoking or soul-bruising discussions of super hearing abilities... Klippel allows the assessment of different types of distortion commonly found in speakers (see attached):
www.klippel.de/fileadmin/klippel/Files/Know_How/Literature/Papers/Klippel_Nonlinearity_Poster.pdf
Nonlinearities in suspension, motor, and inductance all contribute unique distortion. The suspension compliance leads to harmonic distortion, the Bl and L(x) nonlinearities show up as IMD-like distributions. Here are the spectra of each distortion type in the test at -39dB, the individual nonlinearities are each offset by 50dB and all is smoothed to aid the eye.

You can see the simple harmonics and sensitivity to low frequency of the compliance nonlinearities, and the IMD-like distributions caused by the Bl and L(x) nonlinearities.

PKane's DISTORT is also useful for simulating distortion, if you want to explore. I think the Klippel pdf linked above poster and the actual nonlinearities explained is the really interesting part though.

 

[Mr. Haelscheir]

View attachment 418472

Longer answer...
In each of these studies, you will see a large number of outliers. We expect about 10 people in this population to hear combined distortion at -69dB, but in fact we see 305 people who appear to have maxed out the test.

View attachment 418484

Either an unusually large population have outlier-great hearing, or they have good analysis tools.

It's easy to discern -69dB by looking at the spectra:
View attachment 418485
I can do no better than -51dB on this test, and it is a bit exhausting, boredom being part of it. Some of these help at higher volume, which is a bit self-limiting. Each person's ear will affect the result depending on the distortion added, Klippel offers just a few modes, each person will have their own combinations they are particularly sensitive to.

More interesting than ego-stoking or soul-bruising discussions of super hearing abilities... Klippel allows the assessment of different types of distortion commonly found in speakers (see attached):
www.klippel.de/fileadmin/klippel/Files/Know_How/Literature/Papers/Klippel_Nonlinearity_Poster.pdf
Nonlinearities in suspension, motor, and inductance all contribute unique distortion. The suspension compliance leads to harmonic distortion, the Bl and L(x) nonlinearities show up as IMD-like distributions. Here are the spectra of each distortion type in the test at -39dB, the individual nonlinearities are each offset by 50dB and all is smoothed to aid the eye.

View attachment 418488
You can see the simple harmonics and sensitivity to low frequency of the compliance nonlinearities, and the IMD-like distributions caused by the Bl and L(x) nonlinearities.

PKane's DISTORT is also useful for simulating distortion, if you want to explore. I think the Klippel pdf linked above poster and the actual nonlinearities explained is the really interesting part though.

Yeah. It's hard to say how many folks used an FFT viewer to the end of the test or used notch filters to enhance audibility of the distortion. I am otherwise not sure how one would cheat with the music samples, not to say that certain people actually are. I had in https://www.audiosciencereview.com/...pared-to-8680-participants.27664/post-2069281 (post #48) noted the phenomenon of hearing other resonant tones that typically stay put until after a certain frequency of the main tone, said tones not showing up as distortion harmonics in my low-distortion in-ear mic measurements.

 

[MAB]

Yeah. It's hard to say how many folks used an FFT viewer to the end of the test or used notch filters to enhance audibility of the distortion. I am otherwise not sure how one would cheat with the music samples, not to say that certain people actually are. I had in https://www.audiosciencereview.com/...pared-to-8680-participants.27664/post-2069281 (post #48) noted the phenomenon of hearing other resonant tones that typically stay put until after a certain frequency of the main tone, said tones not showing up as distortion harmonics in my low-distortion in-ear mic measurements.

My guess is most.
Audiopeople get wrapped up on their ability to hear distortion, I guess they think of themselves as Bene Gesserit truthsayers. Lots of long-winded stories about listener fatigue due to an odd harmonic, out of place, in an obscure work.

 

[Tell]

with music on HD800:
View attachment 162037

Is this really true? I'm personally at around -36dB on the dual tone test which is a bit better that average, and then even worse with the music test, can't remember exactly but around -24dB. Don't get me wrong dude, I do respect you a lot and what you contribute here, I just can't fathom how anyone can do that top score when the average is so much lower? Sure some of them are untrained, but I've been into audio since I was barely talking so I'm not really untrained in any way but still I'm just above average.

 

[solderdude]

Why would I fake the test ? (real time spectrum analysis does not work at this test anyway)
It was about 3.5 years ago and even then there were 400 participants that also aced the test.

Besides see post # 18 and post #20 who also aced the test so I am hardly the only one here.

I did not do the 2 tone test, only the music test.
The tone tests are not really relevant to music listening and are more interesting for audibility thresholds but as you can see from the bell curve such tests are only meaningful when properly done at the correct level with gear that is beyond question and trained listeners when it is to find thresholds.

I have no idea why a majority of people that took the test don't score well.
Could be:
Level of training (knowing what to listen for)
Not listening loud enough
Gear not up to snuff (a LOT of headphones have high distortion in the lows)

 

[HaveMeterWillTravel]

Yeah. It's hard to say how many folks used an FFT viewer to the end of the test or used notch filters to enhance audibility of the distortion. I am otherwise not sure how one would cheat with the music samples, not to say that certain people actually are. I had in https://www.audiosciencereview.com/...pared-to-8680-participants.27664/post-2069281 (post #48) noted the phenomenon of hearing other resonant tones that typically stay put until after a certain frequency of the main tone, said tones not showing up as distortion harmonics in my low-distortion in-ear mic measurements.

In many things, "playing by the rules" is a requirement for something to work. In the infancy of audio related social media (*) that soon became readily apparent.
Local meetups started to happen which was interesting to say the least. Now, I have always been fascinated in edge cases in most things. But it was essential that the all the necessary rules of the test be explicit established and not just "assumed" by different individuals.


(*) I am unable to remember when rec.audio was founded on Usenet. It was there when I discovered it in the late '80s IIRC.