• 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!

Distortion Listening Test

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
45,612
Likes
252,459
Location
Seattle Area
Ran into this interesting listening test by Klippel with regards to how much distortion you can hear.

This is the link: http://www.klippel.de/listeningtest/lt/

Just casually wanting to see the test, I went with the defaults and for some reason, the levels were exceptionally low. I had to set my headphone level to max and only test with a sealed headphone to even hear it.

On second try, clicking on the volume check provided much, much louder signal but I did not run this test.

I also forgot to select a speaker model, going with "--". Not sure what that represents. There is interesting set of choices in the drop down.

Anyway, was taking the test and recording a snapshot from time to time and it all of a sudden quit on me saying I had run enough. This is my snapshot:

1563778724406.png


The thing is ruthless when you miss, sending you back good number of steps. Once I got a sense of it, I was making good progress until it terminated me at -30 dB. Strangely it says this was my results:

1563778805383.png


-6 dB??? I went to -27 and possibly -30.

Also strange to see a larger number of people (313) detecting -45 dB but not higher distortions above that???

Fascinating that 1921 people could not get past the highest level of 18 dB!!! We have a lot of deaf audiophiles and seemingly, speaker engineers.

I am not very clear what the Real Speaker anchor is doing in there. Is that typical distortion simulated for a speaker? If so, I am able to hear distortion of -30 dB below that?

Here is the only bit on it in the Klippel paper:

1563778967327.png


Anyway, take the test and take a snapshot as you go along as the summary results seems wrong. Don't be shy about the outcome. You won't be judged unless your first name is Thomas, or Sal....
 
Let's break the ice here - first with music, using my Philips SBC-HP820 headphones connected directly to a Lenovo Laptop analog output
HearingTest-FullRange-snap.PNG

HearingTest-FullRange-result.PNG


Then I tried the "simpler" 2-tone test (ahem)
HearingTest-2tone-snap.PNG


HearingTest-2tone-result.PNG

Guess I'm tone-deaf. How can this test be so hard?
 
Looks like the place it puts you is a geometric mean of the failed trials. For me, -6 dB was the middle one so it picked that. Can't make any sense out of why that is appropriate especially when one takes the test without practice.
 
This was my first time doing any audio related hearing tests. I guess the reason might be because I already know my hearing isn't that good in comparsion to my age :D.
It seems my intuition was right. For better or for worse :D
Kippel_distortion_test_result1.png

Here is my first try. Unfortunately I forgot to take a snapshot of the answers I chose, but I do think my first wrong answer was around the "Real Speaker" test.

Kippel_distortion_test_result2.png

And here my second try on the test. Same settings as my first try. Seems like I got used to it? I don't know but it felt more like guessing to me.

Kippel_distortion_test_answers2.png

This time I remembered to take a snapshot of my answers.
 
Looks like I fell in the -18db with my Adam A5X and did better with my headphones down to -45dB before I started losing my hearing and rampaging :D. granted I always tend to play my speaker at lower level so I wouldn't bother the rest of the home.

These are the results with my AKG K7xx with the rampaging

1563812891531.png
 
Yes, training (repeating the test) will definitely help and is recommended for listening tests.
I did this test a while ago and I remember that it was harder than I thought it would be. It felt like being trained for one type of degradation (i.e. AAC) is not necessarily helpful when evaluating other types of distorsion (loudspeakers in this case). Took me a couple tries to figure out where to put my attention. Otherwise I tend to focus on audio coding impairments (and I was on bluetooth headphones when I tried this, my bad).

Not only training is recommended, but discussing what we hear between us is fine too. That's how they would do it at the research center in Ottawa. It's not like you're going to influence people's opinion, you're just educating each other, the test is still blind in the end. I'd be curious to know what helps people make their picks in these tests.
 
I don't understand the final results. I also suspect different kinds of distortion are being used though I'm not sure. I missed the third one. Yet next time through I got further down. And then my final correct choice was -39 db with a miss at -42 db. And they put me at -12 db? That was my 2nd miss which is apparently where they peg it.

I selected full range speaker and was using T. Chapman. Not the Bl(x) version.

Now I was listening over speakers with a ceiling fan on. After the second miss, I turned off the ceiling fan which seemed to help a lot even though any noise from it was not really audible. I suppose it is the ceiling fan doppler distortion (discovered by Ray) that was throwing me off. I also started pausing for 3 minutes every third choice to rest my ears after the 2nd miss.
 
1563821048540.png


Audio Technica ATH-ADG1X (open gaming headset) connected to my MB analog out.

Thoughts...

- those tests are _tiring_ - was planning to go on with a Sennheiser and then my speakers, but that is enough for today... and I probably stopped before I should have - but enough is enough for today.

- @Blumlein 88 have some PC fans and a (relatively quiet) 3D printer running in the background -they didn't seem to hurt much but I can imagine being much more disturbed by a regular or oscillating noise. I don't really know what I should be looking for or focusing on in this test, I just go by gut feeling "as in what sounds cleanest?", maybe looking for a kind of disagreeable vibrato in the voice when it sustained a note? What did you listen for?

- I love Tracy Chapman's music and found her recording to be decent back then, but I don't understand how it can be used as reference these days. In a way, it is the Lenna (https://en.wikipedia.org/wiki/Lenna ) of audio. I can see someone publishing a paper about TC's recording as being among the best in the early 90ies, others quoting that paper and using is as a standard, but I just find it OKish these days.

- as a side note, the AT gaming headset seems quite good for a gaming headset - I bought it on a whim with a nice Prime Day discount, so I have had it for a few days only but it survived a quick subjective comparison with my HD600 (caveat: I am definitely not a headphone guy)
 
Going to take the test now with IEM.

Is it HD or IMD?

Also, if you select another type of speakers you can choose other samples, like speech for microspeaker1. What does that mean?
 
Last edited:
Results
 

Attachments

  • Screenshot_20190722-225427_Chrome.jpg
    Screenshot_20190722-225427_Chrome.jpg
    226.6 KB · Views: 427
Thoughts...

- those tests are _tiring_ - was planning to go on with a Sennheiser and then my speakers, but that is enough for today... and I probably stopped before I should have - but enough is enough for today.
Definitely not a fun test.
 
Guys you must be kidding.

You can clearly hear noise at the end of the "distorted" sample.
("-" mode)
You should have listened to the fidelity, not look for a tell. :) I only listened to the first few seconds of each sample.
 
Seems like we have an explanation though on why unusually high number of people got the perfect answer!

Our urge to win challenges as males takes over science. :)
 
That's what I did until the -6dB one.
Then I heard the noise and couldn't unhear it. Samples are only 5s.
 
Guys you must be kidding.

You can clearly hear noise at the end of the "distorted" sample.
("-" mode)

Hey, we tried to play fair and listened to the music, not to any side-channel :)
But good job on noticing it, I guess that explains the surprising high tail of the results distribution.
 
Back
Top Bottom