• 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!

Neumann KH 80 DSP Monitor Review

Status
Not open for further replies.
I would not worry about that tiny difference in the bass between Amirs and Neumanns measurements, it might just be that the speaker Neumann measured was used less or more. fs tends to usually drop after some use on bass drivers. or there is just some variance during production of the bass drivers. If you buy really expensive drivers some manufacturers sell them in matched pairs but you will never have the exact same t/s parameters for every driver of the same model.
Neumann matches every monitor (even their previously non-DSP models to which below FAQ refers to) to less than 0,5 dB variance:

Are Neumann loudspeakers pair matched?
Pair matching as a concept comes from passive designs where there is little opportunity to tweak each unit to be exactly the same as every other unit. Therefore two units are chosen that are quite closely matched to each other and then they are shipped as a pair. In some cases one sees active products that lack production trimmers, so these products also benefit from the efforts of pair matching.

In our active systems every monitor passes through a final test procedure where production trimmers are adjusted to compensate for tolerances in the electro-acoustical system (drivers, crossovers, etc.). The target response is flat in free-field conditions as the product is a studio monitor - a measurement device for exactly converting the electrical input signal to acoustical pressure at the listening position. All our monitors are calibrated to this standard therefore every monitor is pair matched against every other monitor of the same type.


From http://www.neumann-kh-line.com/neum...rof-monitoring_knowledge_faqs_general-answers
 
Informal Listening Tests
Listening tests as before were with JBL LSR305P on my left as the anchor/reference and the Neumann KH 80 DSP Monitor on the right. Levels were matched using pink noise.

The first noticeable thing was that the KH 80 was most similar to LSR305P than any other speaker I have tested so far (which is not a lot). With other speakers you would immediately hear strong accentuation or lack of energy with the differential being quite large relative to the JBL. Here, all the spectrum was here.

I was then surprised that I did not prefer the KH 80 considering that it has even smoother response than the LSR305P. There were two issues here:

1. The scale was too small. This may be a visual bias but the KH 80 sounded more like a little computer speaker than a large speaker that the 305P sounds like. It was a more focused sound emanating from a smaller source. This was a much smaller issue than #2 though.

2. The JBL had a bit more "zing" and clarity as a result that was just lacking in KH 80. Focusing on the graphs, I noticed that the 305P has good bit of peaking (relatively speaking) in and around the frequencies that the KH 80 is weak in. Since I was doing my listening tests using Adobe Audition, I pulled up its parametric EQ and dialed up the 1.8 khz area by about 3 dB. That made a big difference and brought the response and signature much closer to that of JBL. The good directivity meant that there was no ill effect to such a boost.

I figured if this worked, I would bring up the bass a bit to better match that of 305P. Big mistake. Sound went to hell and fast. This thing is optimized for what it already can there.

I am super anxious to see what the Olive preference score says about KH 80 versus the JBL 305P. If it ranks the KH 80 higher then we have some work to do to rationalize my informal listening tests. If so, then I need to expand my listening tests to be more formal and in better environment than just my desk.

Conclusions
The Neumann KH 80 DSP Monitor measures quite well and seems to be competently designed.. This may make it a very good fit for pro use in recording and mixing. For hifi listening though as I noted, I still prefer the JBL LSR305P. The JBL projects a larger image, has more bass performance and detail in mid-range. It is also cheaper. So personally I would not use the KH 80 DSP but you have the data to decide for yourself.

Net, net, we have some work to do to better objectively qualify performance of speakers like Neumann KH 80.

Yes subjective listening tests are not very meaningful when simply putting the speakers on a desk in a normal room.
There are so many variables.
First of all, the standard is setting up speakers on an equilateral triangle. This matters a great deal as your HRTF differs a LOT when angling speakers wider or narrower.
Secondly, placing the speakers on stands and without a desk or other reflective surfaces, just free space. Equal height for all speakers tested. And I believe there's some ISO standard that says speakers should be more than a meter away from any walls (though if less than a few meters this will still give you a reflection cancellation dip in the bass).
Thirdly, a big room that is well treated. Studio standards. Treated first reflection points, smooth even response / RT60.
Then, when reverb and reflections are low enough you'll find that all of a sudden speaker distortion matters a LOT and becomes very audible and find those LSR305p are big distortion boxes amongst other things :)
Should such listening conditions materialize I can suggest using a standard EQ for the bass equally applied to all speakers to correct for the specific room and placement.

Btw, a subjective test of many nearfield studio speakers was done in a similar way in this thread: https://www.gearslutz.com/board/high-end/851143-high-end-nearfield-test.html
It's 116 pages long and one doesn't learn much from it so not recommended reading :)
 
Oh I don't know if I'd say "most" people use a subwoofer, even among ASR readers. The sans-sub high pass could be a good idea, though some speakers might benefit more from a specific bass cutoff than others, especially if the bass is not flat as it might unduly affect timbral balance.

But bass aside, it might be better to do listening tests in a different setting. I can see how bass reinforcement from a desk would benefit the 305P and hurt the KH80, or at least not help it as much as the 305P. Not sure if amir is using them right on the desks or on a stand, but presumably there'd still be at least a bit of bass reinforcement.

It may also just be that Amirs prefers the 305P for slight differences in bass bass or directivity, or that it jived better with the music he played. They're both good speakers. I bring back the Revel vs M2, where the prettier graph doesn't always win, although this is the opposite in terms of directivity.
Imho for a listening test or comparison without any EQing done, each loudspeaker must be far from boundary surfaces like walls, desks etc, as otherwise not the flattest in the lower region will sound most neutral, but the one that compensates accidentally those acoustic influence best. For example a close vertical surface like a table/desk creates a bump around 150 Hz, therefore some monitors have an inverted switchable filter that tries to compensate that. https://assets.ctfassets.net/4zjnzn...e0b9dc10804/2004_goldberg_makivirta_varla.pdf

That's also the reason why Prof. Goertz does all his studio monitor listening tests only with his EQ based on spacial averaged measurments around listeners position and mainly in the lower frequencies unless there are problems in the direct sound, here for example the filters he used for his KH80 listening test (blue curve is the response without EQ and red with):

KH80HOER-1024x764.jpg


Of course some people will say that not everyone has the luxury of access to such an EQ, that's why in this case I suggest placing them the recommended "Hifi-style", that is quite far from boundaries.
 
Last edited:
Imho for a listening test or comparison without any EQing done, each loudspeaker must be far from boundary surfaces like walls, desks etc, as otherwise not the flattest in the lower region will sound most neutral, but the one that compensates accidentally those acoustic influence best.

This was my primary concern as well. Different speakers will benefit differently from being near a desk/wall/etc, and the best one won't always win.

Ultimately it's probably to just not put too much weight into what speaker is preferred in the listening comparison, especially as we're just starting out. The listening tests are mainly to contextualize the data in this case. Our best case scenario based on existing data is about 86 percent accuracy, and with one listener without subtantial controls in the listening setup, there's sure to be more variation. I'm sure a trend will emerge over the months/years though.
 
What is great about a good speaker is that you can throw it in any situation and it can sound good. Such is the case with JBL LSR 305P Mark ii. When I first got them, I put them in the middle of the room and was amazed by their sound. Now I test them on the desk (elevated a few inches by the way) and they still sound great. But these other speakers do not.

EQ should be reserved for low frequencies where the room controls the response. Above transition frequencies the speaker must not need help and certainly not in narrowband way as this one did. Yes, we may want to deploy tone control in the form of a target curve but no more.

My goal is to identify speakers that are so good that you can buy them with confidence and regardless of your room situation, find them to sound good regardless. Some of you may have other goals which is fine but just want to convey mine.
 
For example a close vertical surface like a table/desk creates a bump around 150 Hz, therefore some monitors have an inverted switchable filter that tries to compensate that. https://assets.ctfassets.net/4zjnzn...e0b9dc10804/2004_goldberg_makivirta_varla.pdf
Nobody who has the remotest interest in good sound better rely on such statistical heuristics:

1579675933834.png


Much less a Pro that has ready access to recording and analysis software. You must measure your room and speakers in situ below transition frequencies and properly EQ the system. No way do you want to follow such "dumb" rules. It is one thing for Genelec to want to find a filter they can embed their speakers to sell as a feature. It is entirely another matter to follow them there as a consumer and listener.

Longer term I may correct for room effect with speakers in my testing. For now, I hope no one is following such recommendations for good sound. Measure, measure and measure when it comes to low frequencies. And correct for them appropriately.
 
Well, that is the problem then. Calibrating those two measurements techniques and stitching them is always subject to some variation/inaccuracy. The Klippel system was designed precisely to fix these issues.

I don't think you can get better accuracy than within +/-1 dB of any method due to setup, microphone calibration, temperature differences. Do you know how your microphones were calibrated and spec?
 
I don't think you can get better accuracy than within +/-1 dB of any method due to setup, microphone calibration, temperature differences. Do you know how your microphones were calibrated and spec?
Yes, there is slight variation in high frequencies:

----
9.00 kHz: +0.182 dB
11.80 kHz: +0.454 dB
15.17 kHz: +0.566 dB
18.77 kHz: +0.601 dB
19.55 kHz: -1.00 dB
----

I have not yet put the correction in there but hope to do so. In grand scheme of things it is not material.
 
In all the discussion about how the listening test conflicts with the measurements/preference rating, did anyone consider that sighted bias may contribute? I know your ears are definitely better trained than mine, Amir, but I don't think any of us are immune to it.

Personally I always find it challenging to overcome the intuition that small speakers "sound small" compared to larger speakers in listening comparisons.
 
Well, that is why the test is marked as informal. For me, I need to conduct them to build as good as a calibration I can between what I am measuring and hearing. I am sharing it with you all just in case it is of value. The rigor will increase in these tests including fully blind ones in the future once there is appropriate fixture.

If I don't listen, there will be even louder cries. :)
 
I think this is another case of a "real world" sample not "measuring up" to the hand-picked reference standard.

There are lots of possible reasons for this, and speculating endlessly without a second or even a third example of the same speaker does not seem productive, especially when the published spec for KH80 says +/- 2dB in the "flat region" (https://www.neumann.com/homestudio/en/kh-80).

Still, it's pretty obvious that even among the very few speakers Amir has measured so far this KH80 is by far the flattest.

I believe so far the Revel had the highest correlation between the published reference and Amir's measurement. I've read somewhere that Revel takes great care in unit to unit consistency.

When Amir has a large number speakers measured, I think it will show the relative rankings do work well, and The Machine (aka The Kippel) is just fine.
 
In all the discussion about how the listening test conflicts with the measurements/preference rating, did anyone consider that sighted bias may contribute? I know your ears are definitely better trained than mine, Amir, but I don't think any of us are immune to it.

Personally I always find it challenging to overcome the intuition that small speakers "sound small" compared to larger speakers in listening comparisons.

I too wonder about this - is there any research on whether a "bigger" speaker, given similar frequency response, actually sounds "bigger"? I have not, anecdotally, experienced this to be the case, but I know many people have.

I can see how a 50" tower might sound bigger than say, a Sonos One. But I do not think the 305P and Neumann are different enough in size for that to be noticeable as an actual physical phenomenon (aside from things like frequency response and transient response, etc)

Ultimately you hear what you hear. But I am curious if there's any research on this. Lots of people swear by their towers, but I think the tiny devialet reactors are some of the "biggest" sounding speakers I've heard, for instance.
 
What is great about a good speaker is that you can throw it in any situation and it can sound good. Such is the case with JBL LSR 305P Mark ii. When I first got them, I put them in the middle of the room and was amazed by their sound. Now I test them on the desk (elevated a few inches by the way) and they still sound great. But these other speakers do not.

EQ should be reserved for low frequencies where the room controls the response. Above transition frequencies the speaker must not need help and certainly not in narrowband way as this one did. Yes, we may want to deploy tone control in the form of a target curve but no more.

My goal is to identify speakers that are so good that you can buy them with confidence and regardless of your room situation, find them to sound good regardless. Some of you may have other goals which is fine but just want to convey mine.

Seems like a good goal.
I don't think you'll find many speakers that'll do that, at least not in the vertical off-axis, but that's ok.

But can I ask you what size room you're doing your subjective listening in, and does it have a high ceiling, how reverberant is it and is there any treatment?
There couldn't be more of a difference between listening in a "normal" room where you're listening to the performance in the "sound" of your room vs listening in a well treated room where you're transposed into the space of the recording.
In the first vs the second situation different things in a speaker become noticeable and become of importance.
 
I too wonder about this - is there any research on whether a "bigger" speaker, given similar frequency response, actually sounds "bigger"? I have not, anecdotally, experienced this to be the case, but I know many people have.

I can see how a 50" tower might sound bigger than say, a Sonos One. But I do not think the 305P and Neumann are different enough in size for that to be noticeable as an actual physical phenomenon (aside from things like frequency response and transient response, etc)

Ultimately you hear what you hear. But I am curious if there's any research on this. Lots of people swear by their towers, but I think the tiny devialet reactors are some of the "biggest" sounding speakers I've heard, for instance.
I know sighted effect on speakers.

A while ago, I was listening to my LSR308's at night, in a dark room. The right speaker was next to a floor lamp, so had a light shining on it. I could swear to you that the right speaker sounded "Much Brighter". Rationally I knew it wasn't true, but it sure sounded that way to me. It was like looking at one of those optical illusions (https://en.wikipedia.org/wiki/Checker_shadow_illusion).

So yeah, sighted tests. FWIW.
 
But can I ask you what size room you're doing your subjective listening in, and does it have a high ceiling, how reverberant is it and is there any treatment?
It is on one side of a loft in a huge space, overlooking the living room with ceiling that climbs way up. Reverberation time is high due to it being a very large space. It is at the other end and mirror image of where my main system is located.

When I get larger hifi speakers, I plan to test them where my main system is.
 
What is great about a good speaker is that you can throw it in any situation and it can sound good. Such is the case with JBL LSR 305P Mark ii. When I first got them, I put them in the middle of the room and was amazed by their sound. Now I test them on the desk (elevated a few inches by the way) and they still sound great. But these other speakers do not.
I can bet with you that under optimal conditions (so not on a desk without any EQ) the KH80 sounds better in nearfield condition for most listeners as I own the 305, its calculated listening position response is smoother and the predicted results of the KH80 based on Toole's criteria favour the KH80 too.

EQ should be reserved for low frequencies where the room controls the response. Above transition frequencies the speaker must not need help and certainly not in narrowband way as this one did. Yes, we may want to deploy tone control in the form of a target curve but no more.
Where did you see narrowband corrections on above KH80 EQ? A desk is a very close and detrimental disturbance that favours a speaker which has a dip in that frequency region like the 305 do.

My goal is to identify speakers that are so good that you can buy them with confidence and regardless of your room situation, find them to sound good regardless. Some of you may have other goals which is fine but just want to convey mine.
We have all those goal, but testing loudspeaker isn't a simple "monodimensional" thing like testing electronics, as it depends on too many variables.

Nobody who has the remotest interest in good sound better rely on such statistical heuristics:

View attachment 46878

Much less a Pro that has ready access to recording and analysis software. You must measure your room and speakers in situ below transition frequencies and properly EQ the system. No way do you want to follow such "dumb" rules. It is one thing for Genelec to want to find a filter they can embed their speakers to sell as a feature. It is entirely another matter to follow them there as a consumer and listener.

Longer term I may correct for room effect with speakers in my testing. For now, I hope no one is following such recommendations for good sound. Measure, measure and measure when it comes to low frequencies. And correct for them appropriately.

Of course EQ based on measurements is better, but please bear in mind that the paper is from 2004 and nowadays Genelec is one of the few loudspeaker/monitor manfacturers that offers an own microphone based correction hardware.

But still the "dumb" old filter worked quite well in most cases as their average measurement from 69 test samples showed:
1.jpg
 
Yes, there is slight variation in high frequencies:

----
9.00 kHz: +0.182 dB
11.80 kHz: +0.454 dB
15.17 kHz: +0.566 dB
18.77 kHz: +0.601 dB
19.55 kHz: -1.00 dB
----

I have not yet put the correction in there but hope to do so. In grand scheme of things it is not material.

Thanks for that.

Do you know how calibration was made? Does Klippel provide inter-lab testing results for each microphone at accredited laboratories?
 
I know sighted effect on speakers.

A while ago, I was listening to my LSR308's at night, in a dark room. The right speaker was next to a floor lamp, so had a light shining on it. I could swear to you that the right speaker sounded "Much Brighter". Rationally I knew it wasn't true, but it sure sounded that way to me. It was like looking at one of those optical illusions (https://en.wikipedia.org/wiki/Checker_shadow_illusion).

So yeah, sighted tests. FWIW.

Next, you're going to tell me copper cables sound warmer and wood-veneer speakers have earthy tones :)!
 
Where did you see narrowband corrections on above KH80 EQ?
??? I made the narrowband correction:

Since I was doing my listening tests using Adobe Audition, I pulled up its parametric EQ and dialed up the 1.8 khz area by about 3 dB. That made a big difference and brought the response and signature much closer to that of JBL. The good directivity meant that there was no ill effect to such a boost.
 
But still the "dumb" old filter worked quite well in most cases as their average measurement from 69 test samples showed:
Not really. From the paper:
1579678264249.png


One-third octave is not useful for low frequency optimization. It completely mispredicts real room response. From Dr. Toole:

1579678603431.png


1/3 octave is useful for overall target curve (even there 1/6th octave may be more useful) but not in low frequencies. It hides way too much.
 
Status
Not open for further replies.
Back
Top Bottom