Audibility of harmonic distortion, even or odd? - AVS Forum
Forum Jump: 
 
Thread Tools
post #1 of 11 Old 01-29-2014, 12:22 PM - Thread Starter
 
Heinrich S's Avatar
 
Join Date: Aug 2007
Posts: 974
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 29
Hi guys,

Want to know from those more experienced what the audibility thresholds, if there are any, for distortion artifacts?

I've been reading around that some people say -70 dB harmonic distortion isn't audible. Others say -100 dB is inaudible.

Is there a way to ABX harmonic distortion in varying amounts while listening to music?
Heinrich S is offline  
Sponsored Links
Advertisement
 
post #2 of 11 Old 01-29-2014, 12:41 PM
AVS Addicted Member
 
Ratman's Avatar
 
Join Date: May 2002
Location: Collingswood, N.J.
Posts: 14,383
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 130 Post(s)
Liked: 271
Arnyk, FMW, Amirm...... put your (typing) gloves back on! biggrin.gif
67jason likes this.
Ratman is offline  
post #3 of 11 Old 01-29-2014, 01:15 PM
Member
 
Wayne Highwood's Avatar
 
Join Date: Jun 2012
Posts: 165
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 35
My hopelessly general response:
Low order harmonics can be tolerated in copious amounts, due to fundamental masking and such.
High order harmonics are objectionable in relatively small quantities; these are the hallmark of a clipping ss amp, so don't buy an amp you will clip!
Henrich, unlike room acoustics, which deserve your attention and study, this topic is, relatively speaking, off in the academic weeds.

Arny or someone will be by with links to the samples with various amounts and types of distortion for your listening pleasure. Sorry I don't have those handy. (You could use the search function to find them yourself, you know.)
Wayne Highwood is offline  
post #4 of 11 Old 01-29-2014, 02:02 PM
Advanced Member
 
andyc56's Avatar
 
Join Date: Feb 2013
Posts: 572
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 17 Post(s)
Liked: 182
Quote:
Originally Posted by Heinrich S View Post

Is there a way to ABX harmonic distortion in varying amounts while listening to music?

Yes, to a reasonable approximation. Keith Howard wrote a program that imparts distortion to an existing WAV file and creates a new, distorted WAV file as output. It's called "AddDistortion" and can be found on this page. It was created as part of an article in Stereophile (with a rather flawed listening test protocol in my view, but that's another discussion), which can be found here. After running this program on a music WAV file, you can then take the distorted and undistorted files and, together with Foobar2000 and its ABX comparator plugin, do an ABX test between the two.

One aside: in almost all cases, the idea of "listen to music with X percent distortion" doesn't make any sense. That's because, for a given nonlinear input-output characteristic, the distortion with a sine wave input depends on the amplitude of the sine wave. When applied to an ever-changing music signal, the nonlinear characteristic imparts more damage percentage-wise to large signals than to smaller ones. There is at least one exception to this: one of Arny's cohorts produced an op-amp circuit called the "grunge circuit" that actually produced distortion which, for a sine wave input, was independent of the amplitude of the sine wave. This was published in an AES article. Creating this kind of nonlinear characteristic takes special effort though, and Keith Howard's software is not capable of doing that.
andyc56 is offline  
post #5 of 11 Old 01-29-2014, 03:15 PM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 14,203
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 636 Post(s)
Liked: 1124
Quote:
Originally Posted by Heinrich S View Post

Hi guys,

Want to know from those more experienced what the audibility thresholds, if there are any, for distortion artifacts?

I've been reading around that some people say -70 dB harmonic distortion isn't audible. Others say -100 dB is inaudible.

Is there a way to ABX harmonic distortion in varying amounts while listening to music?

Yes.

From the standpoint of the listener, it is easy enough to use FOOBAR2000's ABX plug in to compare prefabricated .wav files with varying amounts of distortions of a wide variety added to them. FOOBAR2000 is a nice little freeware windows digital music player that several have reported good results with on AVS.

This begs the question of how to find appropriate files to download. I can tell you that actual choice of music files to operate on is of the essence. The audibility of distortion varies strongly with different kinds of distortion and different music files.

Contrary to audiophile myth, its not necessarily the clarity or beauty of the recording that makes it easiest to use to hear distortion. JJ tells us that MPEG group found that sometimes fairly ugly sounding music yielded the most sensitive results, and IME this is more of a global truth.

For my own work I worked out various strategies for adding the desired amounts of the desired kinds of distortion using CoolEdit Pro 2.1 (now known as Adobe Audition). Describing these strategies in sufficient detail so that anybody could duplicate them would probably involve a great deal of writing.

The first thing I would do is concatenate a header containing appropriate test tones to allow me to easily measure whether I had done what I planned to do, and not done what I didn't want to do (like change the level or frequency response).

Nonlinear distortion can be added to a music file by multiplying the file by itself to as many times as needed to get the desired order of nonlinearity. Varying proportions of the file in distorted and undistorted form were added to get the desired amounts of distortion. The actual proportions used were determined by experimentation after being estimated by theoretical analysis.

A large collection of short musical selections picked to make the various kinds of distortion most audible were freely available for years on my old www.pcabx.com web site which may still be accessed on the Wayback machine.
arnyk is online now  
post #6 of 11 Old 01-29-2014, 03:27 PM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 14,203
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 636 Post(s)
Liked: 1124
Quote:
Originally Posted by andyc56 View Post

Quote:
Originally Posted by Heinrich S View Post

Is there a way to ABX harmonic distortion in varying amounts while listening to music?

Yes, to a reasonable approximation. Keith Howard wrote a program that imparts distortion to an existing WAV file and creates a new, distorted WAV file as output. It's called "AddDistortion" and can be found on this page. It was created as part of an article in Stereophile (with a rather flawed listening test protocol in my view, but that's another discussion), which can be found here. After running this program on a music WAV file, you can then take the distorted and undistorted files and, together with Foobar2000 and its ABX comparator plugin, do an ABX test between the two.

One aside: in almost all cases, the idea of "listen to music with X percent distortion" doesn't make any sense. That's because, for a given nonlinear input-output characteristic, the distortion with a sine wave input depends on the amplitude of the sine wave. When applied to an ever-changing music signal, the nonlinear characteristic imparts more damage percentage-wise to large signals than to smaller ones. There is at least one exception to this: one of Arny's cohorts produced an op-amp circuit called the "grunge circuit" that actually produced distortion which, for a sine wave input, was independent of the amplitude of the sine wave. This was published in an AES article. Creating this kind of nonlinear characteristic takes special effort though, and Keith Howard's software is not capable of doing that.

+1.

The ABX grundge circuit was simplicity in itself - it used a precision half wave rectifier to create the distortion waveform which was mixed back with the original music in user-adjustable proportions.

It was part of a device called "The Audio Chamber Of Horrors" which was demonstrated at various AES meetings during the 1980s and 1990s.

http://home.provide.net/~djcarlst/abx_coh.htm




The actual grundge circuit is described in detail sufficient for personal recreation here: http://home.provide.net/~djcarlst/abx_grun.htm

andyc56 likes this.
arnyk is online now  
post #7 of 11 Old 01-29-2014, 03:38 PM
Advanced Member
 
andyc56's Avatar
 
Join Date: Feb 2013
Posts: 572
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 17 Post(s)
Liked: 182
Quote:
Originally Posted by arnyk View Post


The ABX grundge circuit was simplicity in itself - it used a precision half wave rectifier to create the distortion waveform which was mixed back with the original music in user-adjustable proportions.

It was part of a device called "The Audio Chamber Of Horrors" which was demonstrated at various AES meetings during the 1980s and 1990s.

http://home.provide.net/~djcarlst/abx_coh.htm

The actual grundge circuit is described in detail sufficient for personal recreation here: http://home.provide.net/~djcarlst/abx_grun.htm


Well, there it be! "Audio Chamber of Horrors" LOL. I like that. smile.gif
andyc56 is offline  
post #8 of 11 Old 01-29-2014, 03:47 PM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 14,203
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 636 Post(s)
Liked: 1124
Quote:
Originally Posted by andyc56 View Post

Quote:
Originally Posted by arnyk View Post


The ABX grundge circuit was simplicity in itself - it used a precision half wave rectifier to create the distortion waveform which was mixed back with the original music in user-adjustable proportions.

It was part of a device called "The Audio Chamber Of Horrors" which was demonstrated at various AES meetings during the 1980s and 1990s.

http://home.provide.net/~djcarlst/abx_coh.htm

The actual grundge circuit is described in detail sufficient for personal recreation here: http://home.provide.net/~djcarlst/abx_grun.htm


Well, there it be! "Audio Chamber of Horrors" LOL. I like that. smile.gif

BTW the half wave rectified signal is composed of declining proportions of even order nonlinear distortion. It is unusual in that the percentage of distortion is constant as the input waveform's amplitude varies.

The SQAM files that the MPEG group developed can be downloaded from here: https://tech.ebu.ch/publications/sqamcd
arnyk is online now  
post #9 of 11 Old 01-29-2014, 09:48 PM - Thread Starter
 
Heinrich S's Avatar
 
Join Date: Aug 2007
Posts: 974
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 0 Post(s)
Liked: 29
Quote:
Originally Posted by arnyk View Post

Yes.

From the standpoint of the listener, it is easy enough to use FOOBAR2000's ABX plug in to compare prefabricated .wav files with varying amounts of distortions of a wide variety added to them. FOOBAR2000 is a nice little freeware windows digital music player that several have reported good results with on AVS.

This begs the question of how to find appropriate files to download. I can tell you that actual choice of music files to operate on is of the essence. The audibility of distortion varies strongly with different kinds of distortion and different music files.

Contrary to audiophile myth, its not necessarily the clarity or beauty of the recording that makes it easiest to use to hear distortion. JJ tells us that MPEG group found that sometimes fairly ugly sounding music yielded the most sensitive results, and IME this is more of a global truth.

For my own work I worked out various strategies for adding the desired amounts of the desired kinds of distortion using CoolEdit Pro 2.1 (now known as Adobe Audition). Describing these strategies in sufficient detail so that anybody could duplicate them would probably involve a great deal of writing.

The first thing I would do is concatenate a header containing appropriate test tones to allow me to easily measure whether I had done what I planned to do, and not done what I didn't want to do (like change the level or frequency response).

Nonlinear distortion can be added to a music file by multiplying the file by itself to as many times as needed to get the desired order of nonlinearity. Varying proportions of the file in distorted and undistorted form were added to get the desired amounts of distortion. The actual proportions used were determined by experimentation after being estimated by theoretical analysis.

A large collection of short musical selections picked to make the various kinds of distortion most audible were freely available for years on my old www.pcabx.com web site which may still be accessed on the Wayback machine.

In your experience what is the lowest amount of any type of distortion that a person can hear before it becomes inaudible while listening to music? -70 dB? -100 dB?

Do you have appropriate test signals I can use for testing purposes? Do I add in a music track and then tell the program to add in distortion to amounts I deem acceptable for hearing, or is the actual test signal a music file that has distortion recorded in it? I would be a great exercise to be able to see what is and isn't audible.
Heinrich S is offline  
post #10 of 11 Old 01-30-2014, 06:33 AM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 14,203
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 636 Post(s)
Liked: 1124
Quote:
Originally Posted by Heinrich S View Post

In your experience what is the lowest amount of any type of distortion that a person can hear before it becomes inaudible while listening to music? -70 dB? -100 dB?

Based on DBTs - about -60 dB. That is a more global number. The instantaneous dynamic range of the human ear is about 60 dB under ideal conditions. If something pure is playing at 0 dB, then anything that is 60 dB or more down is highly unlikely to be heard.
Quote:
Do you have appropriate test signals I can use for testing purposes? Do I add in a music track and then tell the program to add in distortion to amounts I deem acceptable for hearing, or is the actual test signal a music file that has distortion recorded in it?

For ABXing you need two files - the original music and the orgional music with the distortion added. If you are interested in studying thresholds, then you need a number of files with different amounts of distortion added.
Quote:
I would be a great exercise to be able to see what is and isn't audible.

This is a link to a Dropbox account that contains a bunch of files that you can listen to and do your own comparisons:

https://www.dropbox.com/sh/7l2zbwmdp88zmi6/niIfI8IVwA

There are 19 very brief files of piano music.

piano_nlref.wav - is the reference music file. Compare all the rest of the files to this one. It is as recorded from a very clean high bandwidth source.

The 18 other files are named as follows:

piano_2nd_10pct.wav

piano is the name of the family of files
2nd is the order of distortion
10pct is the percentage of distortion that was added.

I recommend starting with the highest order of distortion and largest amount of distortion and working down. IOW start with 5th order distortion and 30 percent and first work down the percentage of distortion, and then select the next lowest order of distortion. Repeat.

Each file is about 1 megabyte.

I do not guarantee that these files will remain available indefinately, so grab them while you can! ;-)

Also: Ethan Winer's files of a similar nature:

http://ethanwiner.com/audibility.html
arnyk is online now  
post #11 of 11 Old 01-31-2014, 01:15 PM
AVS Special Member
 
Ethan Winer's Avatar
 
Join Date: Apr 2003
Location: New Milford, CT, USA
Posts: 5,748
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 7 Post(s)
Liked: 133
Quote:
Originally Posted by Heinrich S View Post

Do you have appropriate test signals I can use for testing purposes? Do I add in a music track and then tell the program to add in distortion to amounts I deem acceptable for hearing, or is the actual test signal a music file that has distortion recorded in it? I would be a great exercise to be able to see what is and isn't audible.

Yes, it's a great exercise, and it's not difficult to create your own test files with basic audio editing software. This will give you a good start without having to bother:

Artifact Audibility Report

I expound on this in more detail in my AES Audio Myths video, in the sections starting around 24:00 and 32:00.

--Ethan

RealTraps - The acoustic treatment experts
Ethan's Audio Expert book

Ethan Winer is offline  
Reply Audio theory, Setup and Chat

User Tag List

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off