Marching Onward re: distortion

[March Audio]

I have to raise another point about these speaker reviews in general. That is regarding the usefulness, relevance and validity of equalising the response.

Equalising the response of a speaker that has good off axis response is perfectly valid and useful. However you need accurate measurement data to base the correction on.

No one at home has this. So unless you are using one of the specific speakers tested here you can't do this accurately.

Secondly you also need the replay software that has this functionality. Not everyone has Roon.

Speakers should have correct responses out of the box. Why are we making excuses for them by saying some can be equalised?

 

[jtwrace]

Speakers should have correct responses out of the box. Why are we making excuses for them by saying some can be equalised?

Agree 100%.

 

[ctrl]

I think there are a few schools of thought on this. One is, no one hears distortion less than 10% below 100Hz, some extend that to the edge of Schroeder ~300Hz.

I think the 10% figure is debateable, but yes we certainly are less sensitive to distortion at low frequencies

Studies on the masking effect on low frequencies show that the 10% hearing threshold should be realistic for distortions in the frequency range below 150Hz.

The graph below shows the masking for a 150Hz masker at 70dB SL (sensation level = dB above the individual threshold of pure tone).
The amazing thing is that HD3 is even slightly better masked than HD2 (at higher masker frequencies the opposite is true) - even HD5 is masked as good or better than HD2.

UPDATE: See the response of @andreasmaaan for a possible explanation.

So at frequencies below 150Hz you can even simply use THD as a (crude) measure for the audibility of the distortion.

Already at 70dB SL masker sound pressure @150Hz, the masking is around 10% for HD2-HD5 (on average among the test participants). At higher sound pressure levels the masking effect should increase significantly.

Source: Importance of low frequency masking

Ignoring the (probably deliberate to enhance the impression of bass) peak at 100Hz the bass extension is extremely limited, F3 about 80Hz and F6 about 60Hz.

Believe you refer, with your statements, to the distortion measurement of @amirm - but this is done in the near field .
The real f3 point of the T5V is about 55Hz and not 80Hz.

Interesting that while the LP6 has significantly (around 15 Hz) deeper bass, Amir perceived the T5V as more bassy although its only approximately 1dB between 70-90 Hz while in the rest the LP6 has more bass when both normalized to a similar listening volume, here the estimated in room responses of both:

A possible reason I can think of is the rear vs. frontal port which can couple differently to the room, some real listening position measurements would shed a light on that.

My guess would be simply the frequency response tuning as such. Especially when, as in the listening situation of @amirm, the loudspeaker is relatively free standing in the room, a lower f3 point is hardly noticed due to the small displacement volume of the woofer.

My experience is also that even small changes in the low frequency response can completely change the perception and evaluation of the low bass.

 

[March Audio]

Studies on the masking effect on low frequencies show that the 10% hearing threshold should be realistic for distortions in the frequency range below 150Hz.

The graph below shows the masking for a 150Hz masker at 70dB SL (sensation level = dB above the individual threshold of pure tone).
The amazing thing is that HD3 is even slightly better masked than HD2 (at higher masker frequencies the opposite is true) - even HD5 is masked as good or better than HD2.

So at frequencies below 150Hz you can even simply use THD as a (crude) measure for the audibility of the distortion.

Already at 70dB SL masker sound pressure @150Hz, the masking is around 10% for HD2-HD5 (on average among the test participants). At higher sound pressure levels the masking effect should increase significantly.


View attachment 97397
Source: Importance of low frequency masking




Believe you refer, with your statements, to the distortion measurement of @amirm - but this is done in the near field .
The real f3 point of the T5V is about 55Hz and not 80Hz.



My guess would be simply the frequency response tuning as such. Especially when, as in the listening situation of @amirm, the loudspeaker is relatively free standing in the room, a lower f3 point is hardly noticed due to the small displacement volume of the woofer.

My experience is also that even small changes in the low frequency response can completely change the perception and evaluation of the low bass.

Again debatable and as you say listener dependant. Also a small woofer is going to increasingly struggle with low frequencies and increasing volume levels.

This meant to be a near field monitor. A nearfield measurement is more relevant.

 

[Helicopter]

Agree 100%.

I have to raise another point about these speaker reviews in general. That is regarding the usefulness and relevance of equalising the response.

Equalising the response of a speaker that has good off axis response is perfectly valid and useful. However you need accurate measurement data to perform the correction. No one at home has this. So unless you are using one of the specific speakers tested here you can't do this.

Secondly you also need the replay software that has this functionality. Not everyone has Roon.

Speakers should have correct responses out of the box. Why are we making excuses for them by saying some can be equalised?

I agree too, but I think even without EQ, 3dB at 5.5kHz isn't a big deal in a $200 Chinese speaker. It's not like studios and audiophiles are going to buy these.

If the $350 iLoud MTM or the $500 German real version of this is a little bright, I will object more.

 

[March Audio]

I agree too, but I think even without EQ, 3dB at 5.5kHz isn't a big deal in a $200 Chinese speaker. It's not like studios and audiophiles are going to buy these.

If the $350 iLoud MTM or the $500 German real version of this is a little bright, I will object more.

I agree with the point about context and price.

Thing is its not a 3dB increase at 5 kHz, its a 3dB increase from 5kHz to 20 kHz. That has a big impact on the sound.

I just can't reconcile the measured response and subjective comments.

 

[Helicopter]

I agree with the point about context and price.

Thing is its not a 3dB increase at 5 kHz, its a 3dB increase from 5kHz to 20 kHz. That has a big impact on the sound.

Ok, to clarify, I agree with this too.

 

[jtwrace]

I agree too, but I think even without EQ, 3dB at 5.5kHz isn't a big deal in a $200 Chinese speaker. It's not like studios and audiophiles are going to buy these.

If the $350 iLoud MTM or the $500 German real version of this is a little bright, I will object more.

IMO, any EQ is a big deal regardless of price. Otherwise, why not EQ them in the anechoic to be "perfect" and publish those results to show what is capable? I think when you start to EQ it's a slippery slope. EQ requirement is very room dependent so I think leaving them "flat" as designed and not publishing an EQ would be best. That's just my opinion as the majority of users do not know how to EQ anyway or even measure to know what to EQ. I found out a long time ago in my profession that if you give people more knobs to play with they will likely screw it up worse than just leaving it. Simple is usually best for the masses.

When electronics are measured on the AP the result isn't changed so why change the speaker?

 

[ctrl]

Speakers should have correct responses out of the box. Why are we making excuses for them by saying some can be equalised?

I agree with you.
It should at least be made clear that the recommendation only applies if you have the possibility to set a self-filter (or similar filter).

On closer inspection, the matter is even more complex. Because the loudspeaker has a very good horizontal directivity, frequency response corrections and their effects can be estimated "very easily" - compared to loudspeakers with directivity problems where it can sometimes be very long until success is achieved (or not at all).
T5V:

 

[ROOSKIE]

I have to raise another point about these speaker reviews in general. That is regarding the usefulness, relevance and validity of equalising the response.

Equalising the response of a speaker that has good off axis response is perfectly valid and useful. However you need accurate measurement data to base the correction on.

No one at home has this. So unless you are using one of the specific speakers tested here you can't do this.

Secondly you also need the replay software that has this functionality. Not everyone has Roon.

Speakers should have correct responses out of the box. Why are we making excuses for them by saying some can be equalised?

Well the PEQ corrections often do work very well.
The Kipple data shows directory errors as welI.
In room measurements will generally show you if any PEQ change did not produce the intended changes.
Really no different in my mind than giving an insensitive speakers good rating with the caveat that it needs a powerful amp, or giving a speaker without great bass a high rating because it is awesome if a subwoofer is in play.
I mean by reading the review the data is there and it is obvious it requires eq at leastvin the highs. If that eq does bring it to life then I want to know.
Really anyone here should at least be performing basic room correction for the bass and upper bass with REW & a mic, so a tweak or two to the basic frequency response is easy.
There is free software and a miniDSP is inexpensive alternatively.
While it takes some learning it also takes some learning to do just about anything worthwhile. I think people will get it.

All that said I do suppose two scores would be better.

 

[leeroy 85032]

I have to raise another point about these speaker reviews in general. That is regarding the usefulness, relevance and validity of equalising the response.

Equalising the response of a speaker that has good off axis response is perfectly valid and useful. However you need accurate measurement data to base the correction on.

No one at home has this. So unless you are using one of the specific speakers tested here you can't do this.

Secondly you also need the replay software that has this functionality. Not everyone has Roon.

Speakers should have correct responses out of the box. Why are we making excuses for them by saying some can be equalised?

it certainly has a "cherry picked" feel to some of the evaluations .. doesn't it?

 

[andreasmaaan]

A plausible explanation - cool!
This would also explain why at 50dB SL it "only" shows the expected "moving together of the listening thresholds" for HD2-HD5.
It would be instructive to have more sources on the subject of "masking in the low bass range".

In the study one can see the effect of "moving together of the listening thresholds" with the 350Hz masker.
View attachment 97411
A broadening of the masking range towards low frequencies is already evident in Zwicker's studies.
View attachment 97413
Would be really interesting to see how it behaves at 100Hz.

If I understand Zicker and Fastl correctly, it can actually be inferred from their generalised masking graph and knowledge of the critical bandwidth at any frequency of interest.

They say (p. 75):

The dependence on masker frequency can be avoided if the critical band rate is used as the abscissa instead of a logarithmic frequency scale.

And then they give formulae for calculating the critical bandwidth per frequency:

EDIT: oops, meant to include this graph too (p. 163):

 

[PeteL]

Equalising the response of a speaker that has good off axis response is perfectly valid and useful. However you need accurate measurement data to base the correction on.

No one at home has this. So unless you are using one of the specific speakers tested here you can't do this.

Some of us have used our ears when using a PEQ with arguably appreciable results, that's one thing, and MANY of us know how to use a measurment mic and DSP. So not no one at home.

 

[Lorenzo74]

I have to raise another point about these speaker reviews in general. That is regarding the usefulness, relevance and validity of equalising the response.

Equalising the response of a speaker that has good off axis response is perfectly valid and useful. However you need accurate measurement data to base the correction on.

No one at home has this. So unless you are using one of the specific speakers tested here you can't do this.

Secondly you also need the replay software that has this functionality. Not everyone has Roon.

Speakers should have correct responses out of the box. Why are we making excuses for them by saying some can be equalised?

Agree as well. It seems subjective listening is contrasting the objective results...
Upper Trebles are more than 3db higher than mids and bass roll off earlier than Kali LP6...

is the Oskar’s invention (AMT) that create that JOY? Maybe...

I’m curious.
best

 

[beagleman]

I have to raise another point about these speaker reviews in general. That is regarding the usefulness, relevance and validity of equalising the response.

Equalising the response of a speaker that has good off axis response is perfectly valid and useful. However you need accurate measurement data to base the correction on.

No one at home has this. So unless you are using one of the specific speakers tested here you can't do this.

Secondly you also need the replay software that has this functionality. Not everyone has Roon.

Speakers should have correct responses out of the box. Why are we making excuses for them by saying some can be equalised?

Not totally disagreeing with you for real. But response that sounds "Right" on the high end, can vary a good bit from person to person.

One persons "A bit bright" can be anothers "Just right" The high end hearing of guys between 20-70 varies a LOT.

I have heard people say a speaker was so bright it annoyed them, and found other to say it was just about right.

 

[KSTR]

I have to raise another point about these speaker reviews in general. That is regarding the usefulness, relevance and validity of equalising the response.

Equalising the response of a speaker that has good off axis response is perfectly valid and useful. However you need accurate measurement data to base the correction on.

No one at home has this. So unless you are using one of the specific speakers tested here you can't do this.

Secondly you also need the replay software that has this functionality. Not everyone has Roon.

Speakers should have correct responses out of the box. Why are we making excuses for them by saying some can be equalised?

IMHO, everyone who is really serious about fidelity has to use DRC, even in acoustically well-designed rooms. Every software player has options for convolver plugins, no big deal. Buy a calibrated UMIK-1 or UMIK-2 for little $ and you're set. The filter kernels also can be created with free software (rephase).
Getting full blown DRC is doable for way less than $100, this way.

 

[March Audio]

Well the PEQ corrections often do work very well.

Really anyone here should at least be performing basic room correction for the bass and upper bass with REW & a mic, so a tweak or two to the basic frequency response is easy.
There is free software and a miniDSP is inexpensive alternatively.
While it takes some learning it also takes some learning to do just about anything worthwhile. I think people will get it.

All that said I do suppose two scores would be better.

Yes I said that EQ is valid and can work if the speaker has good of axis response.

I'm still clear on why it should be necessary to do so, why someone should need to fix a problem, or why a speaker that needs it should be given a free pass.

 

[March Audio]

]

it certainly has a "cherry picked" feel to some of the evaluations .. doesn't it?

I don't think deliberate but I do see inconsistencies.

That comment is meant as constructive criticism.

 

[March Audio]

Not totally disagreeing with you for real. But response that sounds "Right" on the high end, can vary a good bit from person to person.

One persons "A bit bright" can be anothers "Just right" The high end hearing of guys between 20-70 varies a LOT.

I have heard people say a speaker was so bright it annoyed them, and found other to say it was just about right.

Aaahhh you can't bring individual preference into this. You especially shouldn't be accounting for hearing loss.

The testing here is fundamentally in line with technical measures and Toole research where to simply paraphrase a speaker that has a flat on axis response and smooth off axis response will be preferred by the vast majority.

If you bring individual preference into this then anything goes, anything is good because someone somewhere likes it.

 

[March Audio]

Cheap, measures well, active, sounds good, what's not to like. Thank you @amirm .

Its FR only measures well if you EQ it. It has high LF distortion that even with masking taken into account is borderline audible. Certainly don't turn the volume up any further. It has limited LF extension.

Regardless it is certainly cheap and great VFM. So in that context its great.

There are 2 things to consider here, Absolute performance and then in the context of price and market segment.