• 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). There are daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Speakers distortion

OP
Krunok

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,036
Location
Zg, Cro
The RTA shows you the level of your test tone and the levels of harmonics at the measurement position. What it doesn't show is what the level of a test tone at the harmonic frequencies would be. The levels of the harmonics are effected by both the amount of distortion and the response at the measurement position - for example, suppose your mic happened to be sat at a perfect null for 2 kHz. A test tone at 1 kHz would show no 2nd harmonic distortion, regardless of the actual level of 2nd harmonic distortion produced. In-room responses are not flat, so all harmonic levels are effected. For more on that read Temme's paper, you may find a copy online. For more general information on distortion measurement this old B&K application note is good, though it predates sweep measurements.

When I made the last distortion measurement with RTA I was moving mic in a same MMM manner when I was making frequency response measurement. That way, the chances that mic is sitting in a perfect null for 2kHz (and any other tone) are zero, but with single sweep it can certainly happen. For that reason I am expecting RTA distortion measurement when done with MMM much more precise than when done with a single sweep. What is your opinion on that?
 

Floyd Toole

Senior Member
Audio Luminary
Technical Expert
Industry Insider
Joined
Mar 12, 2018
Messages
310
Likes
2,848
If I may say so, can your overview be summarized by "trust your ears":cool: and "distortion figures are not really meaningful":oops:?

Does this means that our own attempt at measuring speaker distortion is a waste of time?:facepalm:
So, are we back at choosing our speakers by DBT in our living room?o_O
Or perhaps should we just enjoy the music that our ears have been calibrated to. ;) I bought my speakers in 1981... and I still like them as long as I don't play the music too loud that is ~<80 dB SPL at my chair location - Anyway, I live in an apartment...

Yes, I said that the numerical distortion figures from measurements are not reliable indicators of whether you will hear anything in music. That you play your music at moderate levels is a benefit for you, making life easier for your speakers.

Yes, the ears are definitive - they are used to test the tests.

You could listen to a slowly swept tone at a high level - but be certain to protect your ears. Musicians earplugs should be standard equipment for any conscientious audiophile to preserve hearing. The good ones just turn the volume down, leaving timbre pretty much alone. I have used them for decades when flying, doing noisy household or hobby things, and at overly loud rock concerts. Etymotic.com has them - they were the inventors. The customer earmold ones are very comfortable to wear for long periods. A good investment as one is issued only one set of ears and they do not repair themselves. Save your hearing for music and movies; don't waste it on lawnmowers, saws, vacuums, unmuffled bikes, etc.

Are measurements a waste of time? That cannot be answered with certainty. It is not probable that you will find high levels of distortion in well designed speakers, but it is possible. Usually defects are sample defects, rather then design defects, and the distortion may appear over a narrow band of frequencies. Older speakers can exhibit "rub and buzz" noises as they drift out of alignment. Poor end of line testing can deliver the problem in a new speaker. These are easily detected by manually pushing on the center of the dust cap (gently) and listening for mechanical sounds, or playing a pure tone at a low frequency within the transducer's range to generate maximum movement. If you find this, a new transducer is in order - or a new loudspeaker if exact replacements are not available.
 
OP
Krunok

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,036
Location
Zg, Cro
Yes, I said that the numerical distortion figures from measurements are not reliable indicators of whether you will hear anything in music. That you play your music at moderate levels is a benefit for you, making life easier for your speakers.

Yes, the ears are definitive - they are used to test the tests.

You could listen to a slowly swept tone at a high level - but be certain to protect your ears. Musicians earplugs should be standard equipment for any conscientious audiophile to preserve hearing. The good ones just turn the volume down, leaving timbre pretty much alone. I have used them for decades when flying, doing noisy household or hobby things, and at overly loud rock concerts. Etymotic.com has them - they were the inventors. The customer earmold ones are very comfortable to wear for long periods. A good investment as one is issued only one set of ears and they do not repair themselves. Save your hearing for music and movies; don't waste it on lawnmowers, saws, vacuums, unmuffled bikes, etc.

Are measurements a waste of time? That cannot be answered with certainty. It is not probable that you will find high levels of distortion in well designed speakers, but it is possible. Usually defects are sample defects, rather then design defects, and the distortion may appear over a narrow band of frequencies. Older speakers can exhibit "rub and buzz" noises as they drift out of alignment. Poor end of line testing can deliver the problem in a new speaker. These are easily detected by manually pushing on the center of the dust cap (gently) and listening for mechanical sounds, or playing a pure tone at a low frequency within the transducer's range to generate maximum movement. If you find this, a new transducer is in order - or a new loudspeaker if exact replacements are not available.

What is in your opinion the main reason why we are not even for a moment fooled that it is a living person singing/talking in our room when listening even to the best recordings being reproduced by our speakers?
 

Floyd Toole

Senior Member
Audio Luminary
Technical Expert
Industry Insider
Joined
Mar 12, 2018
Messages
310
Likes
2,848
What is in your opinion the main reason why we are not even for a moment fooled that it is a living person singing/talking in our room when listening even to the best recordings being reproduced by our speakers?

Starting with stereo is the basic problem - two channels cannot replicate the performance space. Because there are only two channels, all of the images between the speakers, the phantom images, are double-mono amplitude panned creations resulting in acoustical crosstalk at the ears (both speakers + reflections are delivering the same sounds to both ears. There is an octave wide, up to 10 dB deep cancellation dip around 2 kHz in the direct sound for the featured artist in the center. Figures 7.1 and 7.2 in my book describe the horror story. For a credible front soundstage we need at least three channels. Adding a credible ambiance requires more. Stereo is a directionally and spatially deprived medium and no amount of money or tweaking can get around it.
 
OP
Krunok

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,036
Location
Zg, Cro
Starting with stereo is the basic problem - two channels cannot replicate the performance space. Because there are only two channels, all of the images between the speakers, the phantom images, are double-mono amplitude panned creations resulting in acoustical crosstalk at the ears (both speakers + reflections are delivering the same sounds to both ears. There is an octave wide, up to 10 dB deep cancellation dip around 2 kHz in the direct sound for the featured artist in the center. Figures 7.1 and 7.2 in my book describe the horror story. For a credible front soundstage we need at least three channels. Adding a credible ambiance requires more. Stereo is a directionally and spatially deprived medium and no amount of money or tweaking can get around it.

So you're saying that lack of fidelity we are hearing when it comes to reproducing human voices by modern speakers is not related to distortion but to spatial/imaging issue?
 

Floyd Toole

Senior Member
Audio Luminary
Technical Expert
Industry Insider
Joined
Mar 12, 2018
Messages
310
Likes
2,848
So you're saying that lack of fidelity we are hearing when it comes to reproducing human voices by modern speakers is not related to distortion but to spatial/imaging issue?
If you are looking for reasons why stereo does not sound "real", non-linear distortions would be far down my list. There are obvious reasons screaming at us. The ONLY reason we ended up with two channels is due to the limitations of the then current delivery medium, the LP, which has other issues as well. This is 2019 - time to move on, but sadly not likely.
 

Thomas savage

Grand Contributor
The Watchman
Forum Donor
Joined
Feb 24, 2016
Messages
10,224
Likes
16,078
Location
uk, taunton
If you are looking for reasons why stereo does not sound "real", non-linear distortions would be far down my list. There are obvious reasons screaming at us. The ONLY reason we ended up with two channels is due to the limitations of the then current delivery medium, the LP, which has other issues as well. This is 2019 - time to move on, but sadly not likely.
Ironically we are down to one speaker in most homes now..
 

Cosmik

Major Contributor
Joined
Apr 24, 2016
Messages
3,075
Likes
2,158
Location
UK
What is in your opinion the main reason why we are not even for a moment fooled that it is a living person singing/talking in our room when listening even to the best recordings being reproduced by our speakers?
Speak for yourself. :)

If a recording of a voice is dry and uncompressed, and played through a speaker at the right volume, it stands a pretty good chance of fooling you - even the dispersion characteristics may be about right. If the recording inherently has the ambience of a cathedral, is compressed and played back at some unrealistic volume then cognitive dissonance means it will not fool you.

A friend of mine leaves Radio 4 playing in her living room when she goes out. Many's the time A couple of times I have been fooled into thinking that there was a bunch of people in there talking and laughing because Radio 4 plays often have the right ambience to make it sound just right. (And that's just a boombox type stereo)
 
Last edited:

andrew

Active Member
Joined
May 3, 2018
Messages
159
Likes
34
Starting with stereo is the basic problem - two channels cannot replicate the performance space. Because there are only two channels, all of the images between the speakers, the phantom images, are double-mono amplitude panned creations resulting in acoustical crosstalk at the ears (both speakers + reflections are delivering the same sounds to both ears. There is an octave wide, up to 10 dB deep cancellation dip around 2 kHz in the direct sound for the featured artist in the center. Figures 7.1 and 7.2 in my book describe the horror story.

Is this finding the case for all types of speakers (i.e., directional vs wide controlled directivity vs omni-directional) and all types of rooms; i.e., high-damped first reflection points vs no treatment? And are there any ways within the constraints of stereo to address the problem; e.g. via DSP?
 

Floyd Toole

Senior Member
Audio Luminary
Technical Expert
Industry Insider
Joined
Mar 12, 2018
Messages
310
Likes
2,848
Is this finding the case for all types of speakers (i.e., directional vs wide controlled directivity vs omni-directional) and all types of rooms; i.e., high-damped first reflection points vs no treatment? And are there any ways within the constraints of stereo to address the problem; e.g. via DSP?

More reflections (wide dispersion/multidirectional speakers) create confusion for the phantom soundstage images, making them less well defined. This is sometimes approved of for classical, concert hall, music, but perhaps less so for pop, jazz and small ensemble classics. Hard panned L & R images are quite stark without reflections, so some vote for more reflections to soften those images. DSP can modify some things, but not others. The early "room effects" performed in DSP gave it a bad name. They were corny and tasteless for the most part. DSP employed in multichannel upmixing of stereo recordings is a whole different thing, and it can be excellent.

Some people like binaural processing to create an extremely wide soundstage - having nothing to do with how the art was created, merely gratifying a "gee whiz" desire. As I keep on saying, stereo delivery is so deprived of directional and spatial information that people try all kinds of things.

Why not put some effort into supporting multichannel audio? It is the solution we should have had decades ago. Movies started it, and do it, but music folks are stuck in a two-channel rut. The amazing thing to me is that so many consumers think that stereo is superior, that there is a purity to it that multichannel lacks. Pity.
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
17,143
Likes
29,860
Speak for yourself. :)

If a recording of a voice is dry and uncompressed, and played through a speaker at the right volume, it stands a pretty good chance of fooling you - even the dispersion characteristics may be about right. If the recording inherently has the ambience of a cathedral, is compressed and played back at some unrealistic volume then cognitive dissonance means it will not fool you.

A friend of mine leaves Radio 4 playing in her living room when she goes out. Many's the time A couple of times I have been fooled into thinking that there was a bunch of people in there talking and laughing because Radio 4 plays often have the right ambience to make it sound just right. (And that's just a boombox type stereo)
I've had similar experiences. A friend had Maggie MG2s with G.A.S. amp and SS ARC pre fed by Tandberg FM tuner. Local public station played classical piano music each night.

With open front door on his porch it sounded like a piano was just inside on some recordings. Walk thru the door and it was just HiFi. The illusion was strikingly strong.
 

JohnPM

Senior Member
Technical Expert
Joined
Apr 9, 2018
Messages
310
Likes
804
Location
UK
When I made the last distortion measurement with RTA I was moving mic in a same MMM manner when I was making frequency response measurement. That way, the chances that mic is sitting in a perfect null for 2kHz (and any other tone) are zero, but with single sweep it can certainly happen. For that reason I am expecting RTA distortion measurement when done with MMM much more precise than when done with a single sweep. What is your opinion on that?
The mention of a null was just an extreme example to make a point, whatever you do the response is a long way from flat, which is the basic problem when you are trying to compare the levels of harmonics to the level of the fundamental.
 
OP
Krunok

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,036
Location
Zg, Cro
The mention of a null was just an extreme example to make a point, whatever you do the response is a long way from flat, which is the basic problem when you are trying to compare the levels of harmonics to the level of the fundamental.

The response being "a long way from flat" is the environment speakers are working in so I don't have a problem if that will influence in-room measurement. That is the very reason why speakers sound differently in the room vs in anechoic environment so I expect that to be caught by measurement as it is part of the "truth" being measured.
 
OP
Krunok

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,036
Location
Zg, Cro
Speak for yourself. :)

If a recording of a voice is dry and uncompressed, and played through a speaker at the right volume, it stands a pretty good chance of fooling you - even the dispersion characteristics may be about right.

I wouldn't say that. With some recordings it comes pretty close. If I'm not paying attention I may be fooled for a moment but it never fooled me when I was really listening to a recording.

@Blumlein 88 my friend has a piano and from time to time I come to him to listen to him play it. It is a true concert piano and its sound so much dominates the room with energy that I cannot possibly imagine speakers coming even close to it. Mine cerrtainly don't.
 

Cosmik

Major Contributor
Joined
Apr 24, 2016
Messages
3,075
Likes
2,158
Location
UK
The response being "a long way from flat" is the environment speakers are working in so I don't have a problem if that will influence in-room measurement. That is the very reason why speakers sound differently in the room vs in anechoic environment so I expect that to be caught by measurement as it is part of the "truth" being measured.
Until you move a few millimetres and come up with a different figure..?

I am beginning to think that 'amateur' distortion measurements are a waste of time. Assuming the drivers are half decent, the real action is where you avoid egregious distortion through design decisions. e.g. number of 'ways', motion feedback. If you are stretching a driver over a wider frequency range than necessary, or requiring massive displacement without motion feedback, then distortions of various kinds go through the roof. If you make the right design decisions, the results of distortion measurements will become exceedingly boring :).
 
OP
Krunok

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,036
Location
Zg, Cro
Until you move a few millimetres and come up with a different figure..?

As I explained, with RTA moving mic method I'm taking >70 samples within the space of 1m x 50cm x 30cm. That same way I was making frequency response measurement and they were also consistent.

When I repeat distortion measurement I made that way I get pretty much the same figure.
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
17,143
Likes
29,860
snip


@Blumlein 88 my friend has a piano and from time to time I come to him to listen to him play it. It is a true concert piano and its sound so much dominates the room with energy that I cannot possibly imagine speakers coming even close to it. Mine cerrtainly don't.

I'm not unfamiliar with the power of the piano. But listening thru a doorway from outside it was fully believable you'd step inside to find a piano. I heard it several times and for more than just a minute.
 
OP
Krunok

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,036
Location
Zg, Cro
I'm not unfamiliar with the power of the piano. But listening thru a doorway from outside it was fully believable you'd step inside to find a piano. I heard it several times and for more than just a minute.

Sure, speakers can do that sometimes when you're not in the same room. Unfortunately, as soon as you step in the magic is gone.
 

Blumlein 88

Grand Contributor
Forum Donor
Joined
Feb 23, 2016
Messages
17,143
Likes
29,860
I'm wondering if you could include a 2 khz tone at lower level with the 1 khz tone. Then look at how much the 2 khz tone is above the expected result. This way any resonant or nulling effects would be the same for distortion and signal. Letting you obtain a clean relative result. Do the same for each harmonic. You'd need to run it once with the extra tone and once without. You could of course include as many harmonics in the test signal as you wished.
 
OP
Krunok

Krunok

Major Contributor
Joined
Mar 25, 2018
Messages
4,600
Likes
3,036
Location
Zg, Cro
I'm wondering if you could include a 2 khz tone at lower level with the 1 khz tone. Then look at how much the 2 khz tone is above the expected result. This way any resonant or nulling effects would be the same for distortion and signal. Letting you obtain a clean relative result. Do the same for each harmonic. You'd need to run it once with the extra tone and once without. You could of course include as many harmonics in the test signal as you wished.

What would be the point fo that as I'm moving the mic during measurement? Wavelength of 2kHz tone is app 17cm so I would encounter tops and downs many time during my measurement.

Besides, you own the mic too so why don't you try to make some meaurements and see what you will get? :)
 
Top Bottom