Small footprint monitors for the desk: Price/Performance winner?

[dominikz]

measurement via 48000 smoothed cycles device and windowed phase smoothing

In case you are talking about quasi-anechoic (temporally-gated) method of acoustic high-frequency response measurement, it is true that resolution is progressively lost under approx. 1kHz, with very low resolution in the 150-300Hz range, and practically no resolution below that (which is why it is in practice usually spliced with nearfield low-frequency driver measurement).
Exact frequency/resolution relationship will depend on how much distance from reflective surfaces you can achieve - VituixCAD Time Window calculator tool can help you estimate this when preparing the measurement.
However above approx. 1kHz the measurements obtained with the quasi-anechoic method match really well with measurements done in a calibrated anechoic chamber or Klippel NFS - that is assuming good measurement methodology, of course!
Here's one of my own measurement examples (compared to @amirm's NFS and the reference measurement done by the manufacturer):

Example of a quasi-anechoic measurement done by by @napilopez compared to @amirm's NFS (some more examples here):

I personally did several such measurements, extensive comparisons with 3rd party data, as well as extensive listening tests and have yet to meet any significant mismatch which is not stemming from some kind of methodology error.

In my previous posts I explained a few main types of measurement and what they are used for, as well as referred to research on audibility of "zero phase" crossovers in loudspeaker.
You're of course free to provide data or research coming to different conclusions - after all such discussions are at the heart of this forum. We're all here to learn new things and share our own knowledge!

especially thing i am mentioning about

It is still unclear to me what exactly you think is the issue - hard data would probably help; but as far as I understood your previous posts you seem to claim that the issues are not measurable by any standard method. I don't dispute you heard some issue, but I find it very unlikely the issue would be unmeasurable. Any real audible issues are usually trivially measurable.

then there is not sense to discuss any more

To be honest, I thought it was unlikely that we will come to an agreement from the start of this discussion. But many people read this forum so I thought some of them might still find the exchange useful.

Kali is crappy junk. Cheers and many people who are not paid bt them knows that.

As I said in my previous post, everyone is entitled to their opinion - I'm not trying to change your mind and my responses were written in good faith.
Personally I went with a different brand of monitors (initially JBL LSR series, moved to Neumann KH series since), but I still can't see why Kali wouldn't be a solid choice for a lot of people.

 

[Mort]

Ha, I love how even a simple question devolves into a debate about audio measurements.

Under $1k I love the Philharmonic Audio Ceramic Mini for $850.

 

[RoA]

Have you thought about the small Kali LP-UNF?, at only €280 you could also buy the Kali WS - 6.2 subwoofer and be under €1000, add in the IsoAcoustics iso 130 stands , I imagine you’d be hard pushed to find a better desktop setup

Isoacoustic ISO-130 stands

Kali WS - 6.2 subwoofer here

Imho, many cheap active 'studio' monitors suck. Cheap cabinets, cheap drivers, cheap amplifiers ... they hiss. It's around £700/per speaker where it starts to get interesting ie. Genelec's and Adam's (but not the budget range). At least it was so when I last visited my local pro music store with their own, dedicated listening room with instant switching. Anyone thinking of spending 'only' a few hundred is imho far better off with a headphone system.

 

[galileon]

In case you are talking about quasi-anechoic (temporally-gated) method of acoustic high-frequency response measurement, it is true that resolution is progressively lost under approx. 1kHz, with very low resolution in the 150-300Hz range, and practically no resolution below that (which is why it is in practice usually spliced with nearfield low-frequency driver measurement).
Exact frequency/resolution relationship will depend on how much distance from reflective surfaces you can achieve - VituixCAD Time Window calculator tool can help you estimate this when preparing the measurement.
However above approx. 1kHz the measurements obtained with the quasi-anechoic method match really well with measurements done in a calibrated anechoic chamber or Klippel NFS - that is assuming good measurement methodology, of course!
Here's one of my own measurement examples (compared to @amirm's NFS and the reference measurement done by the manufacturer):

Example of a quasi-anechoic measurement done by by @napilopez compared to @amirm's NFS (some more examples here):

I personally did several such measurements, extensive comparisons with 3rd party data, as well as extensive listening tests and have yet to meet any significant mismatch which is not stemming from some kind of methodology error.

In my previous posts I explained a few main types of measurement and what they are used for, as well as referred to research on audibility of "zero phase" crossovers in loudspeaker.
You're of course free to provide data or research coming to different conclusions - after all such discussions are at the heart of this forum. We're all here to learn new things and share our own knowledge!

It is still unclear to me what exactly you think is the issue - hard data would probably help; but as far as I understood your previous posts you seem to claim that the issues are not measurable by any standard method. I don't dispute you heard some issue, but I find it very unlikely the issue would be unmeasurable. Any real audible issues are usually trivially measurable.

To be honest, I thought it was unlikely that we will come to an agreement from the start of this discussion. But many people read this forum so I thought some of them might still find the exchange useful.

As I said in my previous post, everyone is entitled to their opinion - I'm not trying to change your mind and my responses were written in good faith.
Personally I went with a different brand of monitors (initially JBL LSR series, moved to Neumann KH series since), but I still can't see why Kali wouldn't be a solid choice for a lot of people.

Hi Domic,
I am glad U are trying and almost everything U said i agree but -

1. U still referening to minimal phase measurements and i am talking about different behaviour

2. I can make the wooden trashy box with random speakers inside and will measure it via any available for us system and it will measure perfectly! But the feeling after listening will be bad - not so bad as kali is - but will be not enjoyable - thats why i am in the state - measurements ok but ear is the golden. Of course this what i am refrerring to can be measured but not via noise or sweep - it needs modulated in time signal properly catched and analysed

3. If U will have the chance to check thing with the name 'kali' on it put it next to any properly designed speaker ( even old radio speaker ) and listen to cymbals/hihats. Then please let me know if my testimony was right or not.

Thank U

 

[Miguelón]

Are you aware that the frequency response (magnitude+phase) shows the exact same information as the corresponding impulse (or step) response, just presented in a different way? It is one of the crucial features of Fourier (and inverse Fourier) transform in mathematics, which is what is used to create these graphs.

It also means that the 'time-related' performance of a loudspeaker can fully be predicted from its frequency response (as long as both magnitude and phase is measured - which is normally the case). The two domains (time and frequency) are fundamentally connected.

I wrote about this before, with some examples - perhaps you will find it interesting; especially this part:


As you can see, the "steepness" of the step response (which may be what you call "transient response") depends mainly on how high in frequency the system can play.
Since humans can't hear much above 20kHz (and many much less than that) we don't need the 'transients' to be any 'sharper' than what you can achieve with a system low-passed at around 20kHz.
This is because our hearing itself is already low-passed close to 20kHz.
Any reproduced sound higher than the upper limit of your hearing will not make the 'transients' sound any more 'correct' to you - even if it makes the measured response look nicer!

Note also that we can measure much higher frequencies that we can hear - i.e. there's no problem to measure the step response of a system with an ultrasonic response. But we don't usually focus on that because normally there's no audible benefit.
For example, my measurement microphone is calibrated on-axis up to 25kHz, but I hear only up to about 16-17kHz.

Next, you might argue that 'zero-phase' crossovers are critical for accurate sound, but various well-regarded researchers argue that this is not the case in practice (here's a link to the relevant AudioXpress article with quotes from experts in the field).

It might also be worth mentioning here that a single frequency/impulse/step response is not a full descriptor of tonality in a loudspeaker.
Since loudspeakers radiate sound non-homogenously in 3D space, every point around it will have a (more or less) different measured response to one another. This is why we need to look at the full spinorama when evaluating loudspeaker, and not just the on-axis response. We're looking for both 'flat' on-axis and 'good' (even) directivity.
But even the spinorama has limitations - it is designed to only describe tonality and therefore e.g. doesn't say anything about max SPL capability or non-linear distortion. Which is why we have various additional measurements to test for that.

The suite of measurements published for current Kali Audio monitors indicate these are very well designed, good value loudspeakers; assuming they are setup correctly and not expected to play louder than they can handle. Better loudspeakers exist, sure, but at the price these seem to be pretty good!

Is not truly exact: Fourier transform of a function defined on a compact set gives a unique value, but its image is on the complex plane and has two components.

In sound waves, this is interpreted (in the correct coordinates, usually polar representation of the complex plane) as frequency and phase.

So frequency domain single graphic doesn’t content all of time domain function, we need phase domain graphic.

 

[dominikz]

Is not truly exact: Fourier transform of a function defined on a compact set gives a unique value, but its image is on the complex plane and has two components.

In sound waves, this is interpreted (in the correct coordinates, usually polar representation of the complex plane) as frequency and phase.

So frequency domain single graphic doesn’t content all of time domain function, we need phase domain graphic.

That is correct, which is why in the post you quoted I tried to be careful and to indicate that both frequency magnitude and phase responses are needed for equivalency with time-domain plots, please note the text marked in red:

Are you aware that the frequency response (magnitude+phase) shows the exact same information as the corresponding impulse (or step) response, just presented in a different way? It is one of the crucial features of Fourier (and inverse Fourier) transform in mathematics, which is what is used to create these graphs.

It also means that the 'time-related' performance of a loudspeaker can fully be predicted from its frequency response (as long as both magnitude and phase is measured - which is normally the case). The two domains (time and frequency) are fundamentally connected. (...)

I know that in informal use people use "frequency response" to mean only the magnitude part of the total frequency response, which sometimes causes confusion, and which is also why I included this clarification in my post in the first place.
Perhaps the way I formulated the post made these statements easy to miss - sorry for the confusion if that is the case!

1. U still referening to minimal phase measurements and i am talking about different behaviour

Would you be able to clarify what kind of behavior this is? Is there any technical article that you can link to which explains the issue and the methodology used to detect it? Thanks!

Another note: acoustic measurements done with sweeps are not "minimal phase" in general - the same kind of measurement methodology can be used to measure minimum phase and non-minimum phase systems, and often to measure non-linear distortion too (e.g. in REW).

Of course a measurement could be used to determine whether or not the system's (in this case loudspeaker's) response matches minimum phase behavior or not, but most multi-way loudspeakers are not minimum phase systems so I don't see why you would consider the plots I was showing to be minimum phase.

Perhaps what you meant to say was that the plots didn't show the phase response? If so, that part is true, and it is simply because the pictures were formatted that way to highlight something else - but still the source measurement (done with a simple sweep) contains both phase and magnitude responses.

 

[dominikz]

Cheap cabinets, cheap drivers, cheap amplifiers ... they hiss.

Hiss is a real issue, I agree - but not all people are equally sensitive to it. My old JBLs (LSR305 v1) hissed quite a lot.
JBLs also had a very audible mechanical resonance of the amplifier backplate in one of my two units and in general poor unit-to-unit matching (different tweeter sensitivities and different distortion performance of the on-board amplifier).

The Neumanns that replaced them are basically perfect in all of these categories.

However it would be unfair to say that the JBLs sounded bad - they were actually very good, especially considering the price!
Neumanns were something like 5x more expensive and I can absolutely see how the difference in sound would not be worth it to a lot of people. It was worth it to me, though!

 

[bachatero]

Why make it so complicated? Kali LP-6v2 are nice looking, good sounding, and come in at a bargain basement price.

 

[Miguelón]

That is correct, which is why in the post you quoted I tried to be careful and to indicate that both frequency magnitude and phase responses are needed for equivalency with time-domain plots, please note the text marked in red:

I know that in informal use people use "frequency response" to mean only the magnitude part of the total frequency response, which sometimes causes confusion, and which is also why I included this clarification in my post in the first place.
Perhaps the way I formulated the post made these statements easy to miss - sorry for the confusion if that is the case!


Would you be able to clarify what kind of behavior this is? Is there any technical article that you can link to which explains the issue and the methodology used to detect it? Thanks!

Another note: acoustic measurements done with sweeps are not "minimal phase" in general - the same kind of measurement methodology can be used to measure minimum phase and non-minimum phase systems, and often to measure non-linear distortion too (e.g. in REW).

Of course a measurement could be used to determine whether or not the system's (in this case loudspeaker's) response matches minimum phase behavior or not, but most multi-way loudspeakers are not minimum phase systems so I don't see why you would consider the plots I was showing to be minimum phase.

Perhaps what you meant to say was that the plots didn't show the phase response? If so, that part is true, and it is simply because the pictures were formatted that way to highlight something else - but still the source measurement (done with a simple sweep) contains both phase and magnitude responses.

Sorry, I jumped some lines. You mentioned both components of the Fourier analysis.

Which Neumanns were you talking about? KH 120 ii?

For some reason I didn’t like them, despite the high accuracy of their measurements. Maybe I should have given them more time at home to get used to their room response, or they irradiate a lot at my living room (only made dip switches manipulation, not MA-1 calibration)

 

[dominikz]

Sorry, I jumped some lines. You mentioned both components of the Fourier analysis.

Which Neumanns were you talking about? KH 120 ii?

For some reason I didn’t like them, despite the high accuracy of their measurements. Maybe I should have given them more time at home to get used to their room response, or they irradiate a lot at my living room (only made dip switches manipulation, not MA-1 calibration)

No, I have the older KH120A, which is fully analogue and without the on-board DSP.
I use mine in nearfield on my desktop (at about 80cm distance), have them integrated to a small sub (some in room measurements without and with the sub here), and I use 3-bands of PEQ on my RME audio interface to correct the main room resonances in the bass.

I remember initially I thought the speakers were a bit bright, but very soon I got used to their sound and now I absolutely love them!

 

[Miguelón]

No, I have the older KH120A, which is fully analogue and without the on-board DSP.
I use mine in nearfield on my desktop (at about 80cm distance), have them integrated to a small sub (some in room measurements without and with the sub here), and I use 3-bands of PEQ on my RME audio interface to correct the main room resonances in the bass.

I remember initially I thought the speakers were a bit bright, but very soon I got used to their sound and now I absolutely love them!

Interesting! Finally I decided to get Genelecs 8030 which are not very different on measurements, but also love them. Attending for a sub to complete the spectrum.

KH 120 ii sounded also a point brilliant to me, but surely is because Genelecs have some minor deeps around 2-3 kHz in room response and also around 10 kHz axial. Or other reasons, subjectively sounded better to me.

Which I found very interesting is the correction kit with the mic and the software, but the total was to expensive for my budget, given that Genelecs 8030C and 7050 sub were priced only 200€ above than Neumanns KH 120 ii plus M1-A (MA-1?) mic

 

[dominikz]

Genelecs 8030C and 7050 sub were priced only 200€ above than Neumanns KH 120 ii plus M1-A (MA-1?) mic

I'd probably make the same choice as you in this case.
Having a sub brings several benefits (lower extension, higher SPLs, more flexibility in battling SBIR), and you could still do room correction upstream (e.g. in PC, DSP-enabled audio interface or a dedicated DSP box).
I'm sure the Genelecs sound great!

 

[thewas]

Interesting! Finally I decided to get Genelecs 8030 which are not very different on measurements, but also love them. Attending for a sub to complete the spectrum.

KH 120 ii sounded also a point brilliant to me, but surely is because Genelecs have some minor deeps around 2-3 kHz in room response and also around 10 kHz axial. Or other reasons, subjectively sounded better to me.

Which I found very interesting is the correction kit with the mic and the software, but the total was to expensive for my budget, given that Genelecs 8030C and 7050 sub were priced only 200€ above than Neumanns KH 120 ii plus M1-A (MA-1?) mic

I would also prefer the 8030+7050 due to the reasons @dominikz wrote above plus also the Genelec has rather a dip than a slight peak in the 2-4 kHz presence region:

Above plot generated by the great https://www.spinorama.org/

 

[Miguelón]

I would also prefer the 8030+7050 due to the reasons @dominikz wrote above plus also the Genelec has rather a dip than a slight peak in the 2-4 kHz presence region:
View attachment 390544
Above plot generated by the great https://www.spinorama.org/

I knew the dip, was also referenced by Amir in his review of the 8030C. I’m going to purchase a WiiM Ultra and rise this region, giving that Amir told good results with PEQ without affecting rest of highs performance.

 

[thewas]

I knew the dip, was also referenced by Amir in his review of the 8030C. I’m going to purchase a WiiM Ultra and rise this region, giving that Amir told good results with PEQ without affecting rest of highs performance.

Personally I wouldn't rise it in that region as I prefer rather a small dip then bump there, but that depends also on personal preference, room acoustics, listening distance and songs, so test and decide yourself.

 

[Miguelón]

Speaking about (very) small footprint speakers, IMHO Genelecs 8020D have a great sound and super small size: always can be reinforced by a discrete 7040 sub under the desk…

I have a doubt: we purchased a pair of 8020, and saw that Genelec have a “twin” equivalent home speaker called G Two: same electroacoustics, just RCA input and no gain knob.

In an obscure article about audiophile, I red that “joining too much analogue volume knobs can lead to a dirty signal, better go to a setup with just one master volume”.

Is true that claim? If I have a volume knob in my audio interface (surely an attenuator) and a gain knob in my active monitors the signal will be affected? Is better to go to Genelec G Two which has no volume controller?

 

[dominikz]

In an obscure article about audiophile, I red that “joining too much analogue volume knobs can lead to a dirty signal, better go to a setup with just one master volume”.

Is true that claim? If I have a volume knob in my audio interface (surely an attenuator) and a gain knob in my active monitors the signal will be affected? Is better to go to Genelec G Two which has no volume controller?

The claim might be a misunderstanding of the principles of gain staging in audio systems.
Having some kind of level control on both your audio interface and your monitors is a good thing in my book because it allows you to optimize the analog signal level between them. This can also be helpful when integrating a subwoofer, or if you want to control the maximum level of the system. Personally I see no drawback, as long at the implementation is reasonable (and I'm pretty sure Genelec does it right ).
In my case the audio interface level control is used as a system volume control, so I set the sensitivity on the monitors such that the 0dBFS on my audio interface gives the highest volume I expect to ever need - that makes the full range of the audio interface volume control usable, and it also decreases the risk of deafening myself with the accidental (but unavoidable) playback at full-scale.
If you work in audio production these controls could also help you calibrate towards a known level reference (e.g. using K-system).

 

[blueone]

Hiss is a real issue, I agree - but not all people are equally sensitive to it. My old JBLs (LSR305 v1) hissed quite a lot.

Good point. While I'm disgusted with Audioengine's marketing nonsense lately, neither one of our Audioengine desktop speaker systems (A5 and A2) hiss at all. Complete silence. And I agree, the older LSR305s did hiss, but I haven't heard a pair lately. In a desktop system hiss is completely unacceptable. IMO, hiss should be the first decision in the desktop speaker decision tree.

 

[Miguelón]

The claim might be a misunderstanding of the principles of gain staging in audio systems.
Having some kind of level control on both your audio interface and your monitors is a good thing in my book because it allows you to optimize the analog signal level between them. This can also be helpful when integrating a subwoofer, or if you want to control the maximum level of the system. Personally I see no drawback, as long at the implementation is reasonable (and I'm pretty sure Genelec does it right ).
In my case the audio interface level control is used as a system volume control, so I set the sensitivity on the monitors such that the 0dBFS on my audio interface gives the highest volume I expect to ever need - that makes the full range of the audio interface volume control usable, and it also decreases the risk of deafening myself with the accidental (but unavoidable) playback at full-scale.
If you work in audio production these controls could also help you calibrate towards a known level reference (e.g. using K-system).

Thanks for your comment!

In Genelecs 8020D, the SPL max is 100dB, but the minimum gain is +6dB. Since my audio interface provides 4 Volt (balanced connection) is impossible to achieve 0 dBFS without clipping…

I find this a little bit annoying, it demands a third stage like a monitor control to reduce sensitivity.

At this moment I have both 8020D and G Two at home, and the latest is way more simpler to control volume: is 10dB less sensitive than its professional brother (apart from that, they share same electro acoustic elements).

Despite the benefit of a balanced XLR input, the 8020D doesn’t perform as well as the G Two, should keep the interface analogue volume at 50% (digital at 100%) or to get full analogue volume then digital at 50%…

All that with one single speaker!

Because I love simplicity and didn’t perceived any interference or ground loop in my setup, I will probably keep the G Two and connect to the WiiM Ultra directly by RCA.

I suppose numbers are right: Focusrite Scarlett 4 volt balanced TRS => TS to RCA cable => Genelec G Two = 2 volt unbalanced signal because one of the wires is grounded.

So WiiM Ultra 2 volt unbalanced => RCA to RCA cable => Genelec G Two = 2 volt signal and similar signal strength as the above path.

Am I correct? I forgive something?

POST EDIT: I forgot a third possible path, WiiM Ultra => RCA to XLR cable => Genelec 8020D. Should be 6 dB weaker than the balanced signal by grounding, so it will take the 8020D 6 dB less while preserving full range, and having the benefits of 4 dB more in gain if I play weak recordings…

 

[dominikz]

In Genelecs 8020D, the SPL max is 100dB, but the minimum gain is +6dB. Since my audio interface provides 4 Volt (balanced connection) is impossible to achieve 0 dBFS without clipping…

Note that 4Vrms is about 14,2dBu.
Looking at this article, Genelec speakers reference the input sensitivity as input level in dBu required to achieve 100 dB SPL at 1m - in case of 8020D the same level is also defined as a maximum short-term output.
That means that with the input sensitivity set to minimum (i.e. +6 dBu sensitivity setting) you would start to get clipping when levels hit close to -8 dBFS.

I was unable to find the reference levels for sensitivity of G Two so can't help with that calculation - though I do see it has a switch to reduce input sensitivity by 10dB.