• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Audibility of Harmonic distortion - a blind test proposal

pkane

Major Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
1,235
Likes
1,713
Location
North-East
#1
Soliciting input on a blind test to determine audibility of distortions. Please comment!

In a measurements-focused forum, the question of whether a specific level of distortion is audible or not is the key to bridging the divide to the subjective opinion.
  • At what level does THD or THD+N become audible?
  • Do even harmonics sound better than odd?
  • Can we really hear SINAD of 120dB vs 100dB?
  • Are harmonics further removed from the fundamental more audible?
  • What type of music does THD affect more?
  • Do tubes sound better than SS?
Would it be useful to run an on-line blind test that might help answer this question? A track or two with various combinations of harmonics applied to see what everyone prefers and at what level is the difference audible. Say three files, same track, but with THD at 1%, 0.1%, 0.01% and 0.001%.

Any interest? Thoughts on what music to use? Levels of THD to test? Number of files?

I can reduce or remove odd or even harmonics, or just change any or all harmonics arbitrarily in the custom selector. Here's an example of a generated 47dB SINAD distortion:

1574295435020.png


This is similar to, for example, a tube Erato M pre-amp recently measured by @WolfX-700 :
1574295214646.png
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
8,181
Likes
5,316
Location
Riverview FL
#2
Show the result on two pure tones input.

Say 440 and 659.26 - call it 660 for ease of analysis.

A and E, a fifth musical interval.

(or post a copy to play with)
 
Last edited:

restorer-john

Major Contributor
Joined
Mar 1, 2018
Messages
3,730
Likes
7,290
Location
Gold Coast, Queensland, Australia
#3
Would it be useful to run an on-line blind test that might help answer this question? A track or two with various combinations of harmonics applied to see what everyone prefers and at what level is the difference audible. Say three files, same track, but with THD at 1%, 0.1%, 0.01% and 0.001%.

Any interest? Thoughts on what music to use? Levels of THD to test? Number of files?
If you can play the files, they can be captured. And if they can be captured, they can be analyzed, and some people will cheat. They will.
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
8,181
Likes
5,316
Location
Riverview FL
#4

pkane

Major Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
1,235
Likes
1,713
Location
North-East
#5
Show the result on two pure tones input.

Say 440 and 698.46- call it 700 for ease of analysis.

A and F

(or post a copy to play with)
I'll post the app itself soon enough so everyone can play with it :) but it's not ready for prime time quite yet (thinking of adding other kinds of distortion profiles to it).

Sorry, didn't do any error checking on this one, just quickly added the two frequencies. Two tones, same level of harmonics as in the first post (THD+N of -47dB):

1574296841367.png
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
8,181
Likes
5,316
Location
Riverview FL
#7
Ooh...

That got ugly!
 

Tks

Major Contributor
Joined
Apr 1, 2019
Messages
1,098
Likes
1,166
Location
NYC
#8
Amir posted a site a while back that was testing audibility in actual music (there was a flaw which allowed people to fudge the results), but with that taken out.. if indeed those levels of distortion were true to the file being played. I couldn't hear -60db or I think even -50db..

I think Amir said its not a problem if you know what to listen out for. I on the other hand feel ignorance is bliss in that case lol.

But yeah, no one came remotely close to anything insane like -110db or whatever.
 

pkane

Major Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
1,235
Likes
1,713
Location
North-East
#10
Amir posted a site a while back that was testing audibility in actual music (there was a flaw which allowed people to fudge the results), but with that taken out.. if indeed those levels of distortion were true to the file being played. I couldn't hear -60db or I think even -50db..

I think Amir said its not a problem if you know what to listen out for. I on the other hand feel ignorance is bliss in that case lol.

But yeah, no one came remotely close to anything insane like -110db or whatever.
Ah, so someone already has done this. I was looking for that site but didn't find it. Can you post the link?

-50dB is better than 1% THD, so entirely possible that it's not easily audible.
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
8,181
Likes
5,316
Location
Riverview FL
#11
Care to attempt a major triad?

Say 440, 555, and 700?
 

Tks

Major Contributor
Joined
Apr 1, 2019
Messages
1,098
Likes
1,166
Location
NYC
#12
Ah, so someone already has done this. I was looking for that site but didn't find it. Can you post the link?

-50dB is better than 1% THD, so entirely possible that it's not easily audible.
I'll try to find it, I'll PM you if I do, and edit this post so you can grab the link here (and others as well). I'll also have it PM'd for you specifically.

Found it: http://www.klippel.de/listeningtest/lt/
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
8,181
Likes
5,316
Location
Riverview FL
#14
Have to run out. I’ll post this a bit later. Sounds like you want to beta test the software
Just wanted to see how it handled "music" - which this isn't quite, but is a little closer, having simultaneous tones.

Anyway, it does add harmonics, and difference tones, and looks like it adds harmonics of the harmonics and harmonics of the difference tones, so it generates a real mess.

Well done!

---

I think I prefer less distortion.

Build something that removes it from playback.

---

One tone becomes ten (counted)

Two tones become sixty (quick count of what is visible)

Three becomes... uh... 360? (What's the math for this? 10x10x10 or 1,000? No, that doesn't include the difference tones and their harmonics...)

1574298745483.png
 
Last edited:

pkane

Major Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
1,235
Likes
1,713
Location
North-East
#16
Just wanted to see how it handled "music" - which this isn't quite, but is a little closer, having simultaneous tones.

Anyway, it does add harmonics, and difference tones, and looks like it adds harmonics of the harmonics and harmonics of the difference tones, so it generates a real mess.

Well done!

---

I think I prefer less distortion.

Build something that removes it from playback.

---

One tone becomes ten (counted)

Two tones become sixty (quick count of what is visible)

Three becomes... uh... 360? (What's the math for this? 10x10x10 or 1,000? No, that doesn't include the difference tones and their harmonics...)

View attachment 39605
Here you go. Three tones, same harmonic profile as before (ignore the THD/THD+N numbers -- the software doesn't calculate this correctly if there is more than one fundamental):

1574301309372.png
 

PierreV

Senior Member
Joined
Nov 6, 2018
Messages
498
Likes
1,292
#18
I think that, in principle, it is a great idea.

As far as "cheating" is concerned, I don't see the point of cheating in such a test but yes, I guess some people always will for whatever reason their devious minds can conceive. Taking Paul's points, the answer to the questions

1 - At what level does THD or THD+N become audible?
2 - Can we really hear SINAD of 120dB vs 100dB?


certainly helps define the point where spending more, or going for the latest "measurement champ" stops making sense.
note: I think the Harman "how to listen" test can be used to test 1 but I don't remember my threshold and I have already answered NO to 2 (for myself ofc)

3 - Are harmonics further removed from the fundamental more audible?
4 - Do even harmonics sound better than odd?
5 - What type of music does THD affect more?

Are fun things to test, for fun's sake. But they could also help explain the appeal of some typical harmonic sprays (aka house sound) that make people like or dislike stuff such as the PS Audio, Pass etc... sound.

6 - Do tubes sound better than SS?

The mother of all cans of worms: I like it ;)

The only concern I have is that the test system seems the obvious limiting factor in all those. Those tests assume a perfect baseline - how could one test 120dB vs 100 dB on an intrinsically 90dB limited system? How can one test a specific harmonic profile on a system that already has a characteristic harmonic profile?

Still, even with those limitations, I like the idea. Plus, I don't have to write the code and Paul clearly enjoys what he is - brilliantly - doing :D:D
 

Blumlein 88

Major Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
7,886
Likes
9,523
#19
Be aware on that Klippel test page, they peg your score on the 2nd miss. It takes a couple three tries to get the hang of their test method. Then you can get reasonable results. It was discussed in this post and the few that follow.

https://www.audiosciencereview.com/...ds/distortion-listening-test.8152/post-201862

After some familiarization I seemed stuck hearing a difference to around -39 or -42 db (about 1% distortion).
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
8,181
Likes
5,316
Location
Riverview FL
#20
Don't laugh, but I already have
Can't help it.

Some things just elicit a giggle.

Feed it the output of the harmonized triad above, and post the result... no cheating!
 

Similar threads

Top Bottom