Kali Audio IN-8 Studio Monitor Review

[tktran303]

Looks to me like the W-M crossover is out of wack. The the woofer is a solid 3dB-6dB too quiet. The mid-tweeter SPL levels seems matched.

what’s wrong with this thing- woofer, mid, tweeter, or crossover. By design (suboptimal baffle step compensation) or by damage?

Amir, do you know if this thing is using a passive or active crossover? Programmable DSP or fixed/discrete complements?

Could be an easy fix for Kali, or not...

 

[amirm]

Amir, do you know if this thing is using a passive or active crossover?

I assume active because it has three amplifiers.

 

[napilopez]

First of all, thank you again for these reviews and your investment in the Klippel.

Second, I think things are going to get interesting.

I don't think there are too many speakers in the world that will outperform the 305P on these objective measurements. However, I think we also know that while the 305P are certainly enjoyable in most peoples' subjective opinions, there are many many speakers that are more enjoyable.

So ASR will find ourselves having lots of interesting discussions about that sort of thing.

Probably as soon as you publish the first measurements of a speaker that most people would prefer to the 305P, despite having a worse DI.

The 305P were a perfect choice for a first set of measurements, btw. Cheap, relatively ubiquitous, and extremely well-behaved. Sort of a "model citizen" as far as measurements go.

Interesting discussions? Like if a $5k/pr speaker isn't too bad, but not really equal on sound alone to a $300/pr speaker, but the story, mystique, appearance etc. make it sound better to your sighted then is it really better? Is your musical enjoyment actually real? Even if it isn't and you like it better then it is as real as it can get?

Or maybe the important mods to an LSR305 is to put really fancy veneer on it, give it a different name, and invent a story so it can be enjoyed to the full extent the sound quality really deserves?

I think once we've have a good gamut of good-measuring speakers, things will be interesting. So far we have one "good" speaker and three questionable ones. But unlike dacs and amps, there are many ways a speaker can be good, even within the parameters of "flat-ish dirct sound and smooth off-axis."

As is so often the debate here and elsewhere, I think matters of directivity will come to fore. What's better: A speaker with uber-flat listening window and good directivity, or a speaker with slightly lumpy listening window and super-smooth directivity? How flat does the FR really need to be? What about the occasional bipole, dipole, and omindirectional models?

I also think that even as a group of people who appreciate evidence-based design, we all have little things we tend to focus on more. Some people care a lot about distortion. Others about max SPL. Others wide directivity. Others narrow directivity. Some people want efficiency. I don't mind seeing a bit of a 2Kish boost to account for IAC in stereo listening. Some want a lot of bass extension, which lets not forget, accounts for approximately 30 percent of preference in listening tests all on its own.

The good thing is now we'll have more data than ever to form these opinions, even if we disagree a bit on the interpretation of the data.

 

[Jmudrick]

These guys do measurements and the speakers are DSP driven. There's more to this story. Haven't a clue what it is.

 

[aarons915]

Yeah, the LS50W measures a fair bit flatter than the LS50, and they basically did the same right in the DSP, its seem. (still has the 1.5K dip but disappears off axis, so not hugely audible). Where did you find that study, btw? First time i've seen a spin of the LS50!

I actually stumbled upon it by chance, not even sure why because I'm not really interested in Flat panel type speakers. The Study is called "Quantifying Listener Preference of Flat-Panel Loudspeakers" and the LS50 was used as a control to test against.

 

[KaliAudio_Official]

The measurements that you showed here are NOT in line with what we have for on-axis frequency response, spatial average, or sound power for the IN-8. We would not put a loudspeaker that measured like this on the market. We would like to understand what's going on with the speaker you have.

Your measurements and comments don't line up with what we've experienced with the IN-8, or the reactions we've gotten from others. If the speaker is defective, it is covered under our warranty, and all of our retailers are quite good about replacing product when there's an issue like this. In this case specifically, we would be interested in getting the speaker back to analyze it ourselves. We are more than happy to provide another IN-8, as well as an LP-6 and LP-8 if you'd like to test those as well.

If you reached out to us about this via our online form, email, or social media channels, please accept my apologies for not replying yet. I just spot checked our inboxes and didn't see anything from you.

These speakers have previously been reviewed on this forum, and the experience seems to have been entirely more positive. Full disclosure: we have not had any contact with the person who wrote that review.

As many commenters have pointed out, we do not have FR or CLF data published for the IN-8 yet. Our intention is to have all of our loudspeaker products tested by Pro Sound Testing in Indiana. To be frank, the IN-8 has sold about twice as well as we expected so far, and so we haven't been able to spare a pair to send to them for testing. Perhaps this was hubris on our part, and we apologize for the lack of transparency. It is not our intention to hide the performance of our loudspeakers. Quite the opposite: we prefer to provide 3rd party data for the sake of trustworthiness. To be clear, we will be having this testing done and will publish it on our website as soon as it is complete.

Here is a link to another analysis of the IN-8 where the reviewer did his own measurements: http://www.hifi-forum.de/index.php?action=browseT&forum_id=30&thread=13717&postID=33159#33159

His comments are in German, but you can see the overall frequency response of the speakers. Minus the bumps below 200 HZ, which I write off to the room, these measurements reflect what we would expect as far as the performance of the loudspeaker.

Here you can see one of our own comparisons between the IN-8 (Green) and the LP-8 (Red.) This is an apples-to-apples comparison of these loudspeakers measured in the same place in the same chamber. The reason we're sharing this is that there is a wealth of 3rd party data on the frequency response of the LP-Series (including what's published on our website, linked below,) so you can see the response of the IN-8 in that context. I've also included Stuart Yaniger's review and measurements of the LP-6 below so you can cross-check vs. what we've published.

In the measurements here, you can see some differences between the IN-8 and LP-8. The LP-8 is a bit messier at about 800 Hz. This is a box mode that comes out through the port tube. Because the mid-range handles these frequencies on the IN-8 it doesn't excite the box mode, so you don't see it on the IN-8's response. At 10k, the on axis response shows a diffraction from the baffle edge of the midrange for the IN-8. Looking at the spatial average response (ours is below, but this shows in your measurements as well) you can see that this artifact averages out when off-axis information is introduced.

You can see in these two measurements that the IN-8 and LP-8 are nearly identical below 300 Hz (330 is the xover of the IN-8 woofer) and very close up to 800 Hz. At 800, you see the same port resonance that I mentioned previously. Above that, the two speakers stay very close. Above 4K, octave-to-octave balance is maintained even though the shape of the baffle causes some artifacts in the response.

LP-Series Measurements:
LP-6 vs. LP-8 (Scroll Down https://www.kaliaudio.com/lone-pine-studio-monitors
Stuart Yaniger's (@SIY on this forum) LP-6 Review: https://audioxpress.com/news/focus-on-acoustics-speakers-and-more-in-audioxpress-august-2019

 

[JIW]

Congratulations to the forum member who found the DI curve for early reflections curve is broken.

@Krunok first identified it on the review of the NHTs.

[...]
I have question here: how comes that Early reflections DI doesn't show dip at 4kHz? As value of Listening window at 4kHz is app 73dB and value of Early reflections is app 72dB I would expect Early reflections DI to drop to a value of 1 at 4kHz, but it is not so. How comes?

[...]

View attachment 45674

Then I did some digging to show that the issue is not with Amir's setup but with the Klippel software. Just follow the NHT review thread from the linked post (#27) if you are interested.

 

[amirm]

The measurements that you showed here are NOT in line with what we have for on-axis frequency response, spatial average, or sound power for the IN-8. We would not put a loudspeaker that measured like this on the market.

Welcome to the forum. Do you have any anechoic measurements of these speakers? Without it, it is hard to judge the region between 100 and 300 Hz.

As to this graph:

As I have circled, the cancellation is what I am measuring so we have good correlation there.

If the above low-frequency measurements are performed using near-field, non-anechoic techniques, then as you know that causes a boost in low frequencies due to the baffle, obscuring the issue we I am seeing in my measurements.

Anyway, my measurements are anechoic so until we have a comparable one, it is hard to know if there is an issue here with our work or not.

 

[amirm]

Unfortunately I have returned the speaker that I tested. All is not lost however. If @KaliAudio_Official like to send me one that you have measured, I am happy to remeasure it and report back. There is always the possibility of sample to sample differences so maybe that is in play. If you like to do this, please start a conversation with me (click on my name) and we can take it from there.

Welcome again.

 

[garbulky]

Anyway, my measurements are anechoic so until we have a comparable one, it is hard to know if there is an issue here with our work or not.

I don't mean this in a rude way. Your "anechoic" measurements are obtained by simulation from measurements that are not performed in an actual anechoic environment but in a garage. Not trying to diss your hard work and expense for this undertaking, but I think it's fair to point out that unless I'm wrong, these are thetoretical simulations of what the anechoic measurement should hopefully be.

 

[KaliAudio_Official]

This image is from an anechoic measurement. What you circled is an artifact caused by diffraction from the baffle. You can see on both of our measurements that this artifact disappears once off-axis information is introduced. We'll be in touch about getting you another speaker.

 

[Sancus]

Unfortunately I have returned the speaker that I tested

It might be possible to have the vendor specifically return it to Kali Audio once they've received it. I can check.

 

[Darkweb]

If this performance warrants a decapitaded panther, I expect we will shortly see only parts of panthers lying around (1or 2 legs, a foot, ...). I mean this speaker is very cheap for an active 3-way with 8" woofer and you get what you pay for.

I'm wondering how a much worse measuring speaker for 5 to 10 times the price would be rated. Looking at you, Zu, Devore, horns, ML, ...

There’s simply no way a Devore or Martin Logan sound like cheap clock radios stuffed in boxes though. If they really are “much worse measuring” than these monitors then I’d say the Measurements must be lacking.

 

[Jmudrick]

There’s simply no way a Devore or Martin Logan sound like cheap clock radios stuffed in boxes though. If they really are “much worse measuring” than these monitors then I’d say the Measurements must be lacking.

I have some experience with clock radios and the IN-8 does sound superior.

 

[Arthur Edwards]

@Krunok first identified it on the review of the NHTs.


Then I did some digging to show that the issue is not with Amir's setup but with the Klippel software. Just follow the NHT review thread from the linked post (#27) if you are interested.

Thanks for your work on this very important find. I will follow the posts you produced to get to this point.

 

[napilopez]

Just want to say - it's great whenever a manufacturer comes in to a forum to provide clarity, so I'm happy to see Kali here.

It also highlights the need for manufacturers to provide data in the first place, of course - if Amir had anechoic data to compare to, he could've spotted that something was wrong with his unit or setup - if something is wrong.

I think most of us would be fine with first-party data, because it should be easily replicable, but I do appreciate Kali wanting to have third party data, so hopefully they can provide that soon.

I don't mean this in a rude way. Your "anechoic" measurements are obtained by simulation from measurements that are not performed in an actual anechoic environment but in a garage. Not trying to diss your hard work and expense for this undertaking, but I think it's fair to point out that unless I'm wrong, these are thetoretical simulations of what the anechoic measurement should hopefully be.

Several manufacturers use this system too though, and there's evidence to suggest in at least some respects it's better than anechoic chambers. I don't know if simulation is the right word so much as fancy math revealing the underlying data. It'd be kind of like saying taking a picture with a digital camera isn't capturing the real thing because you need to do some processing of the raw data.

Of course, that doesn't mean there isn't the possibility of error- we see from the ERDI issue that the software isn't infallible, and I've seen misleading measurements from anechoic chambers. But presumably the math itself is down pat.

I think it'll be easier to confirm everything is kosher once Amir has tested a few speakers with extensive measurements available though. Even with the 305p, the best comparison we had was a blurry graph from an older version of the speaker. When I first embarked on speaker measurements I tested the Neumann KH80 since they provide a ton of data to compare against and it measures perfectly flat, plus there was some third party data around.

Before I confirmed I could match Neumann's data closely, I kept on having doubts about whether I was doing measurements right since a few speakers I thought would perform better didn't do so hot. But my measurements matched Neumann's about as well as they could given my gate and splice method, so I had a reference to work with. Hopefully he'll get an extensively measured speaker soon.

 

[Sancus]

When I first embarked on speaker measurements I tested the Neumann KH80 since they provide a ton of data to compare against and it measures perfectly flat, plus there was some third party data around.

Before I confirmed I could match Neumann's data closely, I kept on having doubts about whether I was doing measurements right since a few speakers I thought would perform better didn't do so hot.

I think it'd be awesome if someone provided a KH80 to Amir for testing for that same reason. If nobody else volunteers I may do so myself at some point.

 

[Andreas007]

Hi Amir,

(ha, just saw napilopez post... But I already wrote mine, so here the same opinion)

I think it is really time to measure a speaker like Neumann KH120 or similar.
We need to see how such a monitor (which has very high reputation for being competently designed) performs in your setup.
There would also be all relevant measurements available to compare.

This would allow better assesment of your first measurements.
I don't think your measurements are seriously flawed but since speaker measurement is quite intricate I would feel better to see some sort of a second benchmark (besides the JBL you used for introduction). If you measure a Neumann and results a comparable to those which had been published by the company, I guess discussion about validity of your setup would stop.

At the moment I also have the feeling that a decapitated Pink Panther is a bit too harsh. If studio speakers are already getting decapitated then what to do with "High End" speakers in the wild?

 

[garbulky]

Just want to say - it's great whenever a manufacturer comes in to a forum to provide clarity, so I'm happy to see Kali here.

It also highlights the need for manufacturers to provide data in the first place, of course - if Amir had anechoic data to compare to, he could've spotted that something was wrong with his unit or setup - if something is wrong.

I think most of us would be fine with first-party data, because it should be easily replicable, but I do appreciate Kali wanting to have third party data, so hopefully they can provide that soon.



Several manufacturers use this system too though, and there's evidence to suggest in at least some respects it's better than anechoic chambers. I don't know if simulation is the right word so much as fancy math revealing the underlying data. It'd be kind of like saying taking a picture with a digital camera isn't capturing the real thing because you need to do some processing of the raw data.

Of course, that doesn't mean there isn't the possibility of error- we see from the ERDI issue that the software isn't infallible, and I've seen misleading measurements from anechoic chambers. But presumably the math itself is down pat.

I think it'll be easier to confirm everything is kosher once Amir has tested a few speakers with extensive measurements available though. Even with the 305p, the best comparison we had was a blurry graph from an older version of the speaker. When I first embarked on speaker measurements I tested the Neumann KH80 since they provide a ton of data to compare against and it measures perfectly flat, plus there was some third party data around.

Before I confirmed I could match Neumann's data closely, I kept on having doubts about whether I was doing measurements right since a few speakers I thought would perform better didn't do so hot. But my measurements matched Neumann's about as well as they could given my gate and splice method, so I had a reference to work with. Hopefully he'll get an extensively measured speaker soon.

I'm not arguing the data and I'm not saying that this means the measurement is in error. But I think a more appropriate term is "simulated anechoic measurements". If somebody wants to say well a simulation is just that and not an actual result, (they would be right) but one could point them to the data that shows how close the simulation is in predicting past anechoic measurements. In fact, I think Amir has posted that already.

I also think this would provide some great comparison tests. If Kali provides the true anechoic measurements, and we can compare them to the simulated ones, we can show how well the Kippel system works in his garage, at least with that speaker in simulating an anechoic measurement.

I think if Amir's speaker endeavour takes off, the first thing he would have to do is build up substantial comparisons of simulated anechoic measurements taken of different types of speakers (in his garage) and compare them to real anechoic measurements. The more data there is to compare, the more one can "prove" that Amir's simulated results echo a real testing chamber.
Right now we have evidence that the system works, but imo not enough that it works in Amir's garage. So the more data the better imo.

 

[amirm]

I don't mean this in a rude way. Your "anechoic" measurements are obtained by simulation from measurements that are not performed in an actual anechoic environment but in a garage. Not trying to diss your hard work and expense for this undertaking, but I think it's fair to point out that unless I'm wrong, these are thetoretical simulations of what the anechoic measurement should hopefully be.

I don't know about being rude but you are not understanding the technology or the mathematics/techniques involved.

First, there is no simulation. Real signals are played by the speaker and measured with a microphone. Beyond 1000 kHz time gating is used to completely eliminate reflections so what you get is anechoic by nature. The lower frequencies are corrected using holographic field separation. Dual scans are provided so that it can separate the direct sound from reflected. Finally, near field measurements are translated to far field using Hankel functions of different orders. This has a limitation that doesn't work well above certain frequency. Fortunately the gating is used before that threshold is hit.

Remember, no anechoic chamber is truly anechoic down to 20 Hz. At 20 Hz, the wavelength is whopping 57 feet long. For the chamber to be anechoic in this region, it needs to wedges that are a quarter of this or 14 feet deep. There are none that are that way that I know of. Chambers as such calibrated against free-field response of a speaker measured outdoors from a crane or top of tall building. That calibration is not perfect. Since the chamber is not fully anechoic it means that the placement of the microphone matters as it may land in a room mode/anti-mode. Trial and error is involved to make sure you get true anechoic measurements. Indeed, the Klippel NFS can be used to reduce errors in anechoic chambers:

So both systems have limitations. Neither is a "garage operation." Would you have rated it higher if I had put the measurement system in a white room with people wearing lab coats???

Anyway, I am totally consumed trying to get this effort off the ground while reviewing other bits and manning the forum. Appreciate not having more rocks thrown at me for no good reason.