• 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!

Making sense of Amir's measurements

Mesh

Member
Joined
Nov 11, 2020
Messages
67
Likes
22
I've spent considerable time working at high-end stores with well treated rooms. I've also played the trumpet extensively, largely within big bands and larger orchestras, so I know a good sounding room from one that isn't by ear (granted, more often from where the band/orchestra is sitting, which isn't as well sounding as it is for the audience!). But I know what instruments sound like, and I have spent quite a bit of time in recording studios. I have never really worked with mixing or mastering myself - just listened in and hung around a lot when that work was being done. Aside from being used to what instruments sound like live, I probably have the patience to be a decent second opinion, so that's probably why it turned out that way. However, I'm also a software engineer and researcher, so I know to doubt myself and never get caught thinking I know all there is to know. But I get a lot of the theory and understand what the measurements do.

If I am brutally honest, what I don't get is what measurements are likely to translate into. I usually get by quite well by figuring out what to have a listen to based on tests and reviews here and elsewhere. Most of the stuff, I never buy, and not because it isn't good. I'm just interested in listening and genuinely want to improve my own ability to go from reading tests to listening without getting surprised at what I am hearing. I suspect that many things I hear simply won't be easy to go from measurements to expectations on, and that's fine. What I do want to figure out is what is realistic to suspect hearing, even if the nuances may never be possible to pinpoint. Subjective reviews do their best to try and capture nuances with big words. I don't mind them doing so, but I personally don't even know if we are really speaking the same language when we are describing subjective experiences, so I prefer measurements when possible. And no, coming from the industry before, I never told my customers what they were hearing or should expect. It is better they form their own opinions. Talking about the music is np though, as a good trumpet is a good trumpet even through less than ideal stuff.

I personally have no problem with how I react to peaks or dips in frequency responses, but maybe others do. I know well how e.g. a tad higher levels of certain frequencies can bring out a voice on a particular recording. That can fool many into believing a speaker to be better than something else, when it really is just a coincidence that the recording worked well with that speaker and the preference of the listener (in that room, and so on...).

But what about distortion? Yeah, ofc we can hear it where there is significant amounts of it, but what can I really read out of a plot like this one for example? And please note I am asking about what it would likely translate to when listening, rather than how the graph works.

1608916946880.png


How are these key problem areas showing up when listening?

1608917108030.png


There are more graphs on https://www.audiosciencereview.com/...behringer-b2030p-studio-monitor-review.14719/ for instance, and I have no interest in that particular speaker. I am only using those graphs as examples.

There is just so many different ways to look at distortion, and I have to say I too often just end up going 'oh this one is better/worse than that one' but I have no clue as to how that would manifest itself during listening. Or maybe better worded: What I should be listening for to check if I am picking that up as a problem to me?

I mean, aside from comparing the graphs, what does this one tell me in relation to one on the Behringer monitor above?

1608917464255.png


And what should I be looking for when comparing the last one with this one?

1608917519561.png


I am sure there are a million more questions I have, and maybe others also have, but I wanted to start mainly with my confession as to not knowing how to translate distortion measurements into realistic expectations. I hope I didn't go blind as I didn't find a thread that helped ppl in similar situations to me (in which case, I'd appreciate the pointer to where I can read up more).

Cheers for reading, and hopefully someone can help me out.
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
what should I be looking for when comparing the last one with this one?

They’re the same. Different views though.

Above 100Hz I prefer no harmonics rising above 3% at most with higher SPL — or 10% max. Higher distortion can be painful/fatiguing at loud listening volumes. Below 100Hz 10% is okay. Although it can go up to 20% and maybe even higher in the sub bass.

Some people have stricter personal preferences — they might want nothing above 1% or even 0.5%

Higher distortion will colour what you’re hearing.

You can artificially induce distortion and hear for yourself what your own tolerance level is with a generator or some type of VST effects plugin. Or try a cheap small speaker and push it to its max volume e.g. amazon echo dot (it auto applies compression to the bass at vol. level 6 and above [I think] to limit distortion and protect the drivers).
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
BTW, some of Amir’s earlier graphs can’t be compared with his newer graphs like the one for the Kali IN-8.
 
OP
Mesh

Mesh

Member
Joined
Nov 11, 2020
Messages
67
Likes
22
They’re the same. Different views though.
Yes, I realize that. I am asking why both are there and if anything is possible to translate into expectations as to what may be heard. And I do mean in general rather than that particular speaker.
 

Rock Rabbit

Active Member
Joined
Feb 24, 2019
Messages
230
Likes
174
There's a link between objective measurements and certain audible thresholds. So the graph indicates "warning" zones whereas the hearing threshold is better to detect a problem with a "normal" ear (some statistically average ear).
Measurements and graphics are a very powerful tool to compare gear easily but then is your personal hearing in the equation, maybe the threshold means nothing, even some lightly distorted "warm" sound is better from your perspective...but at least you have bounds and numbers to play with in the decision (especially when there are many variables and options to balance).
Don't take this "thresholds" very seriously...they where measured in a controlled situation...nothing to replicate in a showroom
 
Last edited:

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,204
Likes
16,985
Location
Riverview FL
But what about distortion?

Distortion reduces "clarity". It changes timbre.

Distortion 50dB or more below the original tone may be a nice dividing line between audible and inaudible.

-50dB as percent (as sometimes displayed) is 0.32%

That would (maybe) just be audible with a sine (pure, single frequency) tone.

It will likely be inaudible with most music - where each instrument already produces it own signature of harmonics.

1% to 10% distortion is a range where distortion with music may reportedly become audible - -40dB to -20dB - depending on what the music is...

I have two sets of speakers in this room. One pair distorts earlier and more aggressively than the other. It is only noticable at higher SPL, where the harmonic distortion is much higher.

Here is an 82dB comparison:

Same room, same signal path through the preamp stage.

I measured my little JBL LSR 308 at the SPL they start singing a different song


index.php




and the Martin Logan panels still sounding pure at the same room level...



index.php
 

Sal1950

Grand Contributor
The Chicago Crusher
Forum Donor
Joined
Mar 1, 2016
Messages
14,073
Likes
16,609
Location
Central Fl
 
  • Like
Reactions: PSO

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
why both are there

Sorry, I don’t see why they should not be included.

The best way to find out would be to listen to induced distortion yourself to first establish your own frame of reference. You could do this with your own speakers and find out what levels of distortion you find tolerable.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,375
Likes
234,488
Location
Seattle Area
But what about distortion?
Harmonic distortion measurements as shown here do not have a direct correlation to audibility due to lack of psychoacoustics in them. X% distortion in one situation may be audible but not in another. And of course music presents a different audibility challenge than test tones.

Importantly, people's ability to hear non-linear distortion varies a ton. Some are completely blind to them. At the other extreme trained listeners can hear small impairments at orders of magnitude better than ordinary people/audiophiles.

What we do know is that if distortion is extremely small, we can do away with psychoacoustics and declare it as inaudible and the system transparent. For everyone, every music and every situation. You see this statement in my reviews of high performance headphone amps and DAC. I call this provably transparent. Alas, no speaker comes remotely close to this level.

At the other extreme, we routinely distortion approaching 100% and even going beyond when measuring bookshelf speakers. These are also clearly audible and there is no need for any psychoacoustic analysis. I can easily hear all kinds of noises from crackling, buzzing, rattling, etc. at these extreme levels.

In the middle is the gray area. I wish I could waive my hand and make these THD measurements relative to your music, hearing, and situation but I can't. What I can do is show how two speakers compare and drawing a threshold line that separates the good from not so good. All else being equal, choose the speaker with lower distortion rating.
 
OP
Mesh

Mesh

Member
Joined
Nov 11, 2020
Messages
67
Likes
22
Sorry, I don’t see why they should not be included.

The best way to find out would be to listen to induced distortion yourself to first establish your own frame of reference. You could do this with your own speakers and find out what levels of distortion you find tolerable.
That's not what I was suggesting either. I am asking what each view is trying to illustrate (contribute with) and how that would translate. My issue is not how distortion sounds in general, as I ofc know how I perceive that and how I sometimes perceive it when others are not (and likely how I do not when someone else may). What I am asking for is what indicators in all these different distortion measurements that are more likely to cause what kind of experience? I have never had the fortune to work closely enough with speakers measured with this level of detail to compare the measurements with my sonic experience.
 
OP
Mesh

Mesh

Member
Joined
Nov 11, 2020
Messages
67
Likes
22
Harmonic distortion measurements as shown here do not have a direct correlation to audibility due to lack of psychoacoustics in them. X% distortion in one situation may be audible but not in another. And of course music presents a different audibility challenge than test tones.

Importantly, people's ability to hear non-linear distortion varies a ton. Some are completely blind to them. At the other extreme trained listeners can hear small impairments at orders of magnitude better than ordinary people/audiophiles.
Cheers - this is why I am so interested in learning more about this. And for those that would hear this, what is it that they would perceive? Just 'not sounding right' or is it some other telltale thing?

What we do know is that if distortion is extremely small, we can do away with psychoacoustics and declare it as inaudible and the system transparent. For everyone, every music and every situation. You see this statement in my reviews of high performance headphone amps and DAC. I call this provably transparent. Alas, no speaker comes remotely close to this level.

At the other extreme, we routinely distortion approaching 100% and even going beyond when measuring bookshelf speakers. These are also clearly audible and there is no need for any psychoacoustic analysis. I can easily hear all kinds of noises from crackling, buzzing, rattling, etc. at these extreme levels.

In the middle is the gray area. I wish I could waive my hand and make these THD measurements relative to your music, hearing, and situation but I can't. What I can do is show how two speakers compare and drawing a threshold line that separates the good from not so good. All else being equal, choose the speaker with lower distortion rating.
Fair point, and I am happy you do all the measurements in the first place. They are very helpful. It is indeed the grey zone that has me most interested, and one example I have is that I don't really know if it at all is feasible to tell 2nd harmonic from 3rd harmonic, etc, or even at what levels things may start looking (and thus sounding!) fishy.
 

pma

Major Contributor
Joined
Feb 23, 2019
Messages
4,591
Likes
10,727
Location
Prague
In the middle is the gray area.

Is there any scientific support for this statement, if yes, then post citations, please. Would you also specify what you mean as "middle", which range of nonlinear distortions and what distortion profiles belong to this "gray area in the middle".
 

Ron Texas

Master Contributor
Joined
Jun 10, 2018
Messages
6,078
Likes
8,916
Speaker measurements are the most interesting because there is so much variance among different speakers. Headphones are next. With electronics there is more going on in complex HT controllers and receivers than in single purpose boxes. Let's face it, I don't know who can tell the difference between a system with a cumulative SINAD in the electronics of 80 db and 100 db. Not me for sure.

For starters, the ambient noise in my room was 35 db (A weighted) this morning. That's Christmas day with almost no traffic. I live in a densely developed area less than 1 mile from an interstate highway so there is usually a lot more noise. That's reality and it messes with a lot of stuff. I think @amirm is doing some really brilliant things with his Klippel. This site is having an impact on the audiophile community. I never heard of REW before I joined here and now think the $95 the microphone cost is the most important thing I ever bought in audio.

There are some really smart engineers around here and some equally crazy folk. That's why it's so much fun.
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
I don't really know if it at all is feasible to tell 2nd harmonic from 3rd harmonic.

Why not play with an actual speaker with some actual sounds and measure with a microphone? For example, I know I am quite sensitive to higher order harmonics induced by clipping. How do I know? By actively listening and measuring, comparing what I am hearing with what I have recorded in my own measurements. That's all I can suggest really.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,375
Likes
234,488
Location
Seattle Area
OP
Mesh

Mesh

Member
Joined
Nov 11, 2020
Messages
67
Likes
22
Why not play with actual an speaker with some actual sounds and measure with a microphone? For example, I know I am quite sensitive to higher order harmonics induced by clipping. How do I know? By actively listening and measuring, comparing what I am hearing with what I have recorded in my own measurements. That's all I can suggest really.
I don't even know where to start with a systematic test. I'd love to do one for sure, but I would then need equipment that I feel I can safely submit to such treatment and that I know I am testing in an ideal way so that I don't actually introduce any other variables to it.
Cheers - will do.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
44,375
Likes
234,488
Location
Seattle Area
I don't even know where to start with a systematic test.
A systematic process starts with hearing the distortions in extreme cases. Get a little tube amp, hook it up to a speaker. Crank up the volume and at some point you will hear extreme amount of distortion. Then gradually lower the volume and see if you can still detect the same signature.

Note that it is not clear whether you want to learn to hear and identify distortions. It will lower your detection thresholds and cause you to be bothered with what you used to ignore. :)
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada

You can hear harmonic distortion up to 100% applied to music samples demonstrated in this video.

He's rather fast with the knob turning though!
 
OP
Mesh

Mesh

Member
Joined
Nov 11, 2020
Messages
67
Likes
22
A systematic process starts with hearing the distortions in extreme cases. Get a little tube amp, hook it up to a speaker. Crank up the volume and at some point you will hear extreme amount of distortion. Then gradually lower the volume and see if you can still detect the same signature.

Note that it is not clear whether you want to learn to hear and identify distortions. It will lower your detection thresholds and cause you to be bothered with what you used to ignore. :)
Hehe. I have that problem already! Too many hours spent listening for 'is this thing broken or not' :D

The distortion test was interesting and it seems repeatable for me thus far. If only I could see what in measurements each case actually did (after the answer!), that would be brilliant.
 
Top Bottom