A Blind Test Attempt: Neumann KH310+Sub vs KH120+Sub, and My Thoughts on CEA2034 and Preference Rating

[Pio2001]

Hi Χ Ξ Σ,
Thank you for sharing your work.

Was Dirac On or Off during the listening sessions ?

Did you measure the anechoic response of both speakers (speaker away from walls, microphone 1 meter from the speaker, sweep measurement, impulse response window setup to eliminate first reflections) when they are side to side in order to see how the presence of the other one affects their neutrality ? The diffraction pattern should be completely different from the normal configuration, because with both speakers side to side, we have a virtual extended front side, and the tweeter of the 310 stands in the middle of this large surface, while the tweeter of the 120 is completely off-centered in this surface.

Also, did you measure the average frequency response from the listening position (moving microphone measurement with full range pink periodic noise, RTA) in the configuration of the test ? If there is an audible difference between both speakers, it should be clearly visible here even if they have an identical listening window frequency response.
EDIT : Dirac would cancel these differences.

I think when the spinorama of 2 speakers is very similar (same rating) then other criteria must be used to find the better one. My hint is IMD and in this respect a 3-way always wins over a 2-way. See the multitone measurements by Sound & Recording.

Having the same rating and sounding similar are two different things. Two speakers can have the same rating and sound completely different.
Here, the spinorama show large differences between the KH-120 and the KH-310 in the sound power response (the red dashed curve below the blue one).
This curve strongly affects the in-room result. Here are measurements of my KH-120, in an untreated room, two meters away (directivity indexes are upside down).
We can see that from 1500 to 20000 Hz, the in-room frequency response is identical to the sound power frequency response (if we take into account the treble tilt switch that is set to -1 on my speakers).

Given the large differences between the KH-120 and KH-310 in sound power, there is no need to look further in order to explain the differences.

And that's where an MM measurement would have been useful, because when we look at both spinorama, all we can see is if the curves are even or not, but it is very difficult to see if there is the general slope difference. A 1 dB difference in the slope of the sound power directivity index goes visually unnoticed until we overlay the two graphs, but it can make all the difference between a speaker sounding "shrill" and another sounding "full".

 

[richard12511]

There is no reason for your speaker to reproduce frequencies that does not exist in the recording nor live. That's what the science says. But people believe all kinds of strange things about audio / sound reproduction (see my signature).

Wouldn't those low notes need to produced for the feeling part though? Even if the brain adds the "hearing" part, I wouldn't think it could reproduce the physical force you feel throughout your body with super low notes at high spl.

This song was posted in another thread, but there's a good example of what I'm talking about starting around ~3:46. At high spl, I can physically feel those passing through my body just as much as I can hear them.

 

[Duke]

I've enjoyed your thread, Χ Ξ Σ.

Those waveguide and coaxial design perserve too much high frequency energy... My preferred target is around 1db down per octave for high frequency, but I have not finalized my personal target yet.

I work with constant-directivity waveguides and find that a gently downward-sloping curve (both on-axis and off-axis) is preferred by my beta testers.
This is a somewhat different approach from the Harman target curves.

I think the usual target curves are optimized for cone-n-dome designs, with the target on-axis curve being approximately flat which results in the off-axis energy being gently downward-sloping. The target predicted in-room response is also gently dowward-sloping since it is dominated by the off-axis energy.

With a waveguide speaker imo the target predicted in-room response (for home audio) should also be gently downward-sloping, and because of the similarity between the on-axis and off-axis response, imo the target on-axis response ends up likewise being gently downward-sloping (instead of being "flat").

Imo once a waveguide speaker has been "voiced" with this gently downward-sloping on-axis (and off-axis) curve its tonal balance becomes more relaxing, and we have the added benefit of the waveguide's inherently reduced spectral discrepancy between the direct and "reverberant" sound. Imo spectral consistency between the direct and "reverberant" sound can contribute to natural timbre and long-term fatigue-free listening.

 

[ernestcarl]

But I don't like to listen without correction because the desk bounce not only smears the image but also bring the image down to my chesk level.

You could try raising your stands higher and point the monitors down so that they are angling down — yeah, that sounds redundant.

Although I personally prefer to not have them tilted so that I can walk around and not have the horizontal angle change (stand desk) in relation to my head.

My own desk monitors are pulled much closer-in (desk width narrower) in the very nearfield in order to minimize interference from the monitor and desk. This has the consequence of making slight vertical changes in height cause more obvious shifts in the sound image/character.

 

[Χ Ξ Σ]

In my original in-room measurements, the upper in-room FR is the KH310+sub, the lower is the KH120+sub. I did write "The responses below are my in-room measurements. Both were using Dirac full-range correction. These were taken when the KH120s and KH310s were placed individually before and after the test, not when they placed next to each other, and not when the responses I listened in my blind test." The reason why I posted them was that I wanted to show the FRs I listened to in normal circumstances. Also, I did not measure the side-by-side FR before the test because I did not want to build expectation bias, and I did not do that after the test because I was dealing with an electrical issue and moved everything. However, now I see posting them instead of the uncorrected side-by-side measurements is rather confusing and misleading. Therefore, I asked the KH120 back from my wife and put it right next to the KH310, and did the in-room measurment. I will post the uncorrected side-by-side measurement in the following post.

 

[Χ Ξ Σ]

These are the side-by-side measurement for each channel, recreating the FR I heard during the test. Please note a few things,
1. The UMIK-1 microphone pointed upwards (90deg). Dirac requires 90deg. Therefore, I kept it constant in REW as well. 90deg added more jaggedness than 0deg in the high frequency.
2. Crossover is at 80Hz. Normally I turned the sub volume way up (V14) to provide headroom for DSP. During the test, however, Dirac is turned off, and I turned down the subs' volume by a lot (V9) to prevent the bass from stealing the show. For the purpose of this test, my listening focused on what's about 80Hz.
3. The acoustic center of the KH120 is a few centimeters higher than that of the KH310. while I leveled the mic to the height of the acoustic cent of the KH310, the KH120 measurement naturally shows a roll-off. It is not that important though since my head is not a stationary microphone.

Also why can't I upload a zipped .mdat file?

 

[vavan]

Also why can't I upload a zipped .mdat file?

Now that's something new for me though I didn't do it for a while. Isn't it too big?

 

[Χ Ξ Σ]

Now that's something new for me though I didn't do it for a while. Isn't it too big?

The original was 15MB and the zip was 11MB.

 

[Χ Ξ Σ]

I will try splitting each measurement then.

 

[LeftCoastTim]

Wouldn't those low notes need to produced for the feeling part though? Even if the brain adds the "hearing" part, I wouldn't think it could reproduce the physical force you feel throughout your body with super low notes at high spl.

This song was posted in another thread, but there's a good example of what I'm talking about starting around ~3:46. At high spl, I can physically feel those passing through my body just as much as I can hear them.

You misunderstand me. I'm saying that for piano music, one does not need to reproduce very low frequencies, even though A0 note on the piano seems to be very low frequency.

The OP's tests of KH310 vs KH120 are solely based on piano music, so the low frequency extension of KH310 would have had no impact on the differences he perceived.

 

[teashea]

My desktop 2.2 system was a pair of the Neumann KH120 and a pair of JBL LSR310S subwoofers. I purchased the Neumann KH310s right before Amir tested his unit. The KH310s were not cheap. In order to prove that I could indeed an improvement, I decided to blind test myself and find out whether I truly prefer the more expensive speakers.
View attachment 100806

The Setup

My signal path was MacBook->Motu M2->JBL LSR301S->KH310 or KH120. I connected the M2 with two TRS splitters so that the signal fed into both the LSR310S was stereo. The LSR310S is a typical studio subwoofer that has a built-in high pass filter and balanced in/out to the main speakers. This means when the sub is turned off, the main speakers are silent. I bought two remote controls and connected the receivers to the end of the subs’ power cords, allowing me to switch between the two subs within a few seconds. I tested and verified that one remote only worked with one receiver. I scrambled the identical remotes and marked them A and B using stickers. I piled the stickers off after finishing a set of listening tests, and repeated the process before testing the next set of samples.
View attachment 100807

The LSR310S extends to 27Hz, just barely covers the whole piano range. I could make it extend to 20Hz using Dirac but I did not do that during this test. Adding a sub to each pair of speakers gave them equal bass extension but not equal room resonance. However, the bass quality was not my focus. The point of adding a sub was to see which pair of speakers sound better when only playing above 80Hz.

The KH120s were placed on the outer side and KH310s on the inner side. I eliminated some visual and audio cues that could give it away. I put duct tapes on the subs’ indicator lights so that I wouldn’t know which sub is turned on. The subs would make a popping sound when powered on, therefore, I had to walk outside the room and then use the remotes. This way I couldn’t tell which sub was on, nor could I saw the Neumann indicator lights changing colors.

I was unable to randomize my samples because it was impossible for me to be unaware of which one of the remotes I just pressed. This means my samples had to be ABABABAB and couldn’t be scrambled in to something like ABBABAAB. I verified which speakers that A and B represented after I listened to a group of samples. I know this was imperfect.

I used only classical solo piano pieces as my test tracks for two reasons. A) When playing solo piano music, KH120 often sounded a little shrill, and I couldn’t fix that no matter how I change the placement or use Dirac to add a steep high-frequency tile. B) I needed the music to be as simple as possible. Having multiple instruments in a piece means more spatial information. The more complex the music was, the easier I could identify the speakers by their placements. I did not try to identify which speakers were playing from samples A and B. Rather, I simply put a check next to the samples that I found more pleasing.

The Measurements

@amirm has measured the KH310 a few weeks ago. Guy Layfield from Neumann has posted KH120’s measurements done at Kippel in Germany. Neumann’s website also provides detailed official measurements.
View attachment 100816View attachment 100818

The responses below are my in-room measurements. Both were using Dirac full-range correction. These were taken when the KH120s and KH310s were placed individually before and after the test, not when they placed next to each other, and not when the responses I listened in my blind test.

Can you guess which pair of responses is the KH310+sub, and which pair is the KH120+subs?
View attachment 100819

The Result

I call a single A or B a sample, a pair of A B a trial, and 8 samples or 4 trials a set.

Set One: Nocturne Op. 9/1
B A B A B A B A
X√ X√ X√ √X
A is better on 3 trials, B is better on 1 trial.
A=KH120
B=KH310

Set Two: Prelude Op. 28/1
B A B A B A B A
√X √X √X √X
B is better.
A=KH120
B=KH310

Set Three: Prelude Op. 8/24
B A B A B A B A
√X √X √X √X
B is better.
A=KH120
B=KH310

Set Four: Etude Op. 10/1, Etude S. 144/3
A B A B A B A B
√X √X √X √X
A is better.
A=KH310
B=KH120

Set Five: Ballade Op. 2, Kavierstucke Op. 76/1, Glasswork, Etude No. 9
B A B A B A B A
X√ X√ X√ X√
A is better.
A=KH310
B=KH120

Set Six: Jeux d'eau, M. 30, Nocturne Op. 9/1, Ballade No. 1, Prelude Op. 28/1
B A B A B A B A
√X √X √X √X
B is better.
A=KH120
B=KH310

Total Preference of the KH310:
42/48 samples
21/24 trials
5.25/6 sets

Basically, except for the first 3 trials, I always preferred the sound of the KH310+sub over the KH120+sub in the following 21 trials.

Keep in mind that I couldn’t randomize the samples, I could only randomize the first 2 samples of each set. I think 5/6 is the best number to describe my sccuss rate, i.e., I preferred the KH310 five times out of the six first trials after the two remotes were scrambled.

The Impression

Blind Testing: As I mentioned before, KH120 often had a tiny bit of shrillness when playing classical solo piano pieces. The shrillness of KH120 was inaudible to me when playing symphonies, concertos, black metal, or pop songs, but it irritated me in a very subtle way when playing simple piano notes. KH310 simply does not sound shrill, and that was the main reason I prefer the KH310 in my test.

Sighted Listening: When listening to other genres of music, the gap between these speakers was painfully obvious. Even when the shrillness of the KH120 is not presented, the KH310 had more depth, clarity, authority, smoothness… you name it. Any subjective rhetoric that you are familiar with when switching to a bigger speaker can be used to describe what I heard.

I am not sure what causes the shrillness of the KH120. According to my measurements at my normal listening volume, Both the KH310 and the KH120 produced distortion below the noise floor of my room, so it probably wasn’t due to the distortion. Their frequency responses and directivity are similar, the smaller KH120 might even be better, so it probably wasn’t that either.

The shrillness is probably not caused by the tweeter but the woofer. I found one particularly shrill piano note from the sheet music of a piano piece. The fundamental of that note is at only 349Hz, meaning this note does not activate the tweeter until its sixth harmonic. My theory is that, when holding the tweeter and sub constant, the combination of a dome and a big woofer is better than a small woofer. Quite an obvious explanation, isn’t it? But it isn’t so obvious in the CEA2034 data.
View attachment 100822

The Afterthought

Neumann studio monitors are known for their near-flat on-axis frequency responses and well-controlled directivity behaviors. According to Harman’s research, these two merits, along with the bass extension, are the dominating factors for predicting the preference ratings of the listeners. The KH310 scored 6.2 standalone and 7.6 with sub. We don’t have the data to calculate a score for the KH120, but given its similarity to other Neumann and even Genelec speakers tested by the NFS, I assume the KH120 would score at least 6 standalone and 8 with sub. That should mean the KH120 is as good as the KH310, and the KH120+sub is as good as the KH310+sub, right?

Well, not from my experience.

If I remember correctly, @MZKM posted that the preference rating formula starts to underpredict the actual preference when the rating goes above 6. I believe this is the case here. From what I heard, If the KH120 was rated 6 standalone and 8 with sub, then the KH310 would be 10+. I have yet to find any piece of music that sounds better on the KH120+sub than on the KH310+sub. Even when the shrillness of the KH120 is not present, the KH310 still sounded much fuller and heavier, like if the sound was solid. The KH310 makes the KH120 sound like gas.

I am not throwing shade on Harman’s research. I chose these Neumann studio monitors mainly because their CEA2034 data and presumed preference ratings are quite good. The CEA2034 data available on ASR and other sites have helped me avoid many unworthy options. These measurements allowed me to purchase the KH120 and KH310 without audition and have no regrets. Nonetheless, they have limitations, and it is beyond my understanding of acoustics and speaker design to identify what these limitations are.

I am new here on the forum but throught I would add my thoughts to this thread.

I am the owner of TES Productions. In my studio I have a pair of 120's with a 750 and a pair of 310's. I use them in a nearfield setup (1.2 meters).

I use both in mixing and mastering songs. I find them both useful and informative. They are pretty similar in sound - flat, with little coloration, smooth and even. I use the 310's as my main monitors. I use them about 80 percent of the time.

(I am going to purchase a pair of KH150's the first week of January and will offer comments in the appropriate thread comparing all of them).

 

[Χ Ξ Σ]

I am new here on the forum but throught I would add my thoughts to this thread.

I am the owner of TES Productions. In my studio I have a pair of 120's with a 750 and a pair of 310's. I use them in a nearfield setup (1.2 meters).

I use both in mixing and mastering songs. I find them both useful and informative. They are pretty similar in sound - flat, with little coloration, smooth and even. I use the 310's as my main monitors. I use them about 80 percent of the time.

(I am going to purchase a pair of KH150's the first week of January and will offer comments in the appropriate thread comparing all of them).View attachment 252351

Welcome ASR!

It has been a long time since my original post. My entire setup has changed. I would love to hear your impression of KH310 vs KH150 when you get them.

 

[dasdoing]

But A0 on a concert piano--the string and the soundboard--does not actually produce these frequencies. Only the 3rd harmonic (110Hz) and above are produced by the piano, but your brain makes up the fundamental (27.5Hz) in your head.

it is there, but not audible. and unfortunatly my brain doesn't add those frequencies. always found those notes on a piano horrible, as are the very highest one, too

 

[teashea]

Welcome ASR!

It has been a long time since my original post. My entire setup has changed. I would love to hear your impression of KH310 vs KH150 when you get them.

I will do it.