• 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!

KEF LS60 Room Correction Showdown: REW MMM vs. Wiim Ultra—A Data-Driven Analysis

This can never work? The unit can be positioned everywhere and you still would get a good setting at the listening position?

No. My understanding by reading that comment on the WiiM forum is that the unit needs to be placed at the listening position.
 
Ok, thanks! If that’s accurate, it’s a bit disappointing, as I based my purchase of the LS60 on this information. I understand that my use case is likely in the minority since most users rely on the built-in streaming of the LS60. However, I would expect accurate information in the manual from a company like KEF. Additionally, I really wish they would implement a 10-band PEQ or a room correction feature.
Best is to have accurate info for sure, but would you mind explaining your concern? You want to avoid clipping or an extra volume step?
 
Hello all.

I am relatively new to this forum but I know there is a thread where there iare multiple scientifically documented suggestions NOT to apply these so-called "target curves". For instance, the Harman curve represents the ideal response of a flat speaker with good directional characteristics in a normal room, at far field conditions, therefore it cannot be reproduced with badly performing loudspeakers, i.e. dsp correction could not fix them.

Am I missing or misunderstanding something about the target curves you are talking about in this thread?
 
Hello all.

I am relatively new to this forum but I know there is a thread where there iare multiple scientifically documented suggestions NOT to apply these so-called "target curves". For instance, the Harman curve represents the ideal response of a flat speaker with good directional characteristics in a normal room, at far field conditions, therefore it cannot be reproduced with badly performing loudspeakers, i.e. dsp correction could not fix them.

Am I missing or misunderstanding something about the target curves you are talking about in this thread?
My goal here isn’t to “correct” the speaker itself but rather to address the influence of the room—what we typically refer to as room correction. Even high performing speakers, like the LS60, can’t compensate for room modes, boundary interactions, or other acoustic anomalies caused by the listening environment.

The Harman curve, as you mentioned, is designed to represent the ideal response of a flat, well-performing speaker in a normal room under far-field conditions. Since the LS60 already adheres to this principle by design, my corrections aim solely to “remove the room” and reveal the speaker's true potential. This involves taming peaks and dips caused by room modes—like the ones I encounter at 20 Hz, 60 Hz, and 100–250 Hz in my setup—which would otherwise color the sound and mask its neutrality.

I fully understand the concern about applying target curves to poorly designed speakers, as DSP can’t fix fundamental issues like poor directivity or uneven native frequency response. However, in the case of a high-performing speaker like the LS60, room correction complements its inherent strengths by neutralizing the room's impact, allowing the speaker’s intended performance to shine through.
 
Best is to have accurate info for sure, but would you mind explaining your concern? You want to avoid clipping or an extra volume step?
I'll contact KEF to confirm. I'm using the LS60 with a preamp so I want put the LS60 unity mode, it should ensure that the LS60 does not apply any additional DSP volume attenuation or gain beyond its default calibration.
 
This might not be that important considering that the crossover, with or without the optional phase correction, is implemented in digital domain so any input sample goes through multiple operations before it’s converted to analog for each driver
 
dsp correction could not fix them.

Am I missing or misunderstanding something about the target curves you are talking about in this thread?
There is a big concept here called Shroeder frequency which is defined by the room itself. Below that frequency, room modes tend to dominate, and you can correct them with EQ. The same modes will show up for any speaker you put in the room. This is why it's called room correction and not speaker correction.

Above this frequency, sound is bouncing around more freely in the room (in a manner of speaking) and you need to be more careful with EQ.

There are two main reasons for this:

1) if you correct the sound with a mic in one spot, it is only corrected for that spot. For long wavelengths that's OK because they're huge. For small wavelengths, that is not OK because the "corrected" area is only as big as the wavelength in question. This can be an inch or smaller, so corrections end up making things worse overall.

2) Some speakers radiate sound unevenly depending on the frequency. Especially if you are only correcting with the mic in one spot, your indirect and direct sound end up further out of balance the more EQ you add.

For a speaker like the LS60, with very even radiation, you can use a bit more EQ, but you still need to be careful not to correct narrow frequency ranges or small physical areas in your room - this is true of any speaker.

So TL;DR - you can correct low frequencies always, you can correct higher frequencies sometimes if you use wide (low Q) filters and wider areas for measurement.

If you use VERY wide filters, arguably you can push any speaker toward a target curve, but to an extent that is limited by off-axis unevenness of radiation. I think it's probably ok to use a "treble tilt" on most speakers if you are using an MMM (moving the mic around a wide area) measurement.
 
Thank you @Dako & @kemmler3D for responding. I understand that your primary concern is to remove the room modes below the Shroeder frequency but it seemed to me that the purple curve you have superimposed in the screenshot wasan additional equalization you use, on top of the narrow band attenuation which are intended to mitigate the room modes. If this purple line is just the ideal response you expect from the KEFs, that's fine.

As for the "treble tilt" that kemmler3D mentioned, I think that an a conventional adjustable tone control would be preferable over any fixed attenuation, IMO.
 
Sorry, I haven't read all the replies on this so maybe it's been asked already...

What's the consensus: bad WiiM correction algorithm or bad measurement or both?

If it's a bad algo then using a calibrated measurement mic might not help much.
 
Thank you @Dako & @kemmler3D for responding. I understand that your primary concern is to remove the room modes below the Shroeder frequency but it seemed to me that the purple curve you have superimposed in the screenshot wasan additional equalization you use, on top of the narrow band attenuation which are intended to mitigate the room modes. If this purple line is just the ideal response you expect from the KEFs, that's fine.

As for the "treble tilt" that kemmler3D mentioned, I think that an a conventional adjustable tone control would be preferable over any fixed attenuation, IMO.
So, the purple line does represent the target curve, I think. I haven't played with the WiiM RC tools yet, but in REW you do have a full-range target curve like that, but usually you use settings to tell it to ignore the high frequencies, or to only use wide filters.
 
There is a big concept here called Shroeder frequency which is defined by the room itself. Below that frequency, room modes tend to dominate, and you can correct them with EQ. The same modes will show up for any speaker you put in the room. This is why it's called room correction and not speaker correction.

Above this frequency, sound is bouncing around more freely in the room (in a manner of speaking) and you need to be more careful with EQ.

There are two main reasons for this:

1) if you correct the sound with a mic in one spot, it is only corrected for that spot. For long wavelengths that's OK because they're huge. For small wavelengths, that is not OK because the "corrected" area is only as big as the wavelength in question. This can be an inch or smaller, so corrections end up making things worse overall.

2) Some speakers radiate sound unevenly depending on the frequency. Especially if you are only correcting with the mic in one spot, your indirect and direct sound end up further out of balance the more EQ you add.

For a speaker like the LS60, with very even radiation, you can use a bit more EQ, but you still need to be careful not to correct narrow frequency ranges or small physical areas in your room - this is true of any speaker.

So TL;DR - you can correct low frequencies always, you can correct higher frequencies sometimes if you use wide (low Q) filters and wider areas for measurement.

If you use VERY wide filters, arguably you can push any speaker toward a target curve, but to an extent that is limited by off-axis unevenness of radiation. I think it's probably ok to use a "treble tilt" on most speakers if you are using an MMM (moving the mic around a wide area) measurement.
We also have the psycho acoustic effects described by Toole and others that above the transition frequency we tend to ”hear trough the room”
To a great deal If the speakers dispersion/radiation pattern is welll behaved we tend to hear the on axis sound above the transition frequency ? Our Brian’s will be able to discard some room contributions.
There are exceptions for example if you put the speakers to close to sidewalls the reflected and direct sound is to close in time and the brain thinks they are the same , but EQ can’t fix that either. Or sit to close to a back wall as I do ( some panels behind me helped a bit ).

The consensus seem to be to fix room modes below Schroeder frequency and with some care speaker listening axis sound issues above ? But then you need good speaker measurement from that axis from an anechoic chamber klippel or other methods . As you can’t EQ direct and indirect sound separately. But in case of the LS60 it’s not easy to improve on :) other speakers may have an obvious rise in treble on axis or similar, but then you should really get new speakers.

TLDR .

Below Schroeder frequency the Brian hears the speakers and room as one ( that why you can’t audition subwoofers other than in situ )
Above it’s gets complicated we tend to hear the speakers direct sound but speakers with bad directivity makes weird contributions anyway as the brain likes when direct and reflected sound are similar
 
Follow-Up with Umik-1 Measurements

I revisited the Wiim RC measurements using a iPad Mini (borrowed from a friend) in combination with an Umik-1 microphone The measurement procedure remained the same as before, relying on the REW MMM (Moving Microphone Method) to capture averaged in-room responses.
  • Purple = Wiim RC (20–300 Hz)
  • Orange = REW MMM (20–300 Hz)
  • Red = Harman target curve
    Screenshot 2025-01-11 at 13.45.26.png

From Figure 1, it is clear that Wiim RC (20–300 Hz) (purple) closely aligns with REW MMM (20–300 Hz) (orange) when using the Umik-1. This suggests that the Wiim RC algorithm is highly effective for low-frequency correction without calibration file.

By contrast, in Figure 2, the Wiim RC (20–20 kHz) (blue) introduces significant discrepancies at higher frequencies—most likely because the Wiim software cannot apply a calibration file for the Umik-1, leading to less accurate results above the midrange.
  • Green = Uncorrected
  • Purple = Wiim RC (20–300 Hz)
  • Blue = Wiim RC (20–20 kHz)
  • Orange = REW MMM (20–300 Hz)
  • Red = Harman target curve
Screenshot 2025-01-11 at 13.45.14.png


Conclusion: When Wiim RC is limited to 20–300 Hz and measured with a Umik-1, the outcome is very similar to REW MMM. This indicates Wiim’s room correction can be excellent, as long as the software has accurate measurement data and is restricted to the problem frequencies below ~300 Hz. When Wiim adds support for custom calibration file, the RC solution could become even more robust across the entire bandwidth.

One additional point worth noting is how consistently both Wiim RC (20–300 Hz) and REW MMM (20–300 Hz)maintain a smoother transition between sub-bass and mid-bass. If you look at the 70–150 Hz region, both curves follow a very similar contour, suggesting the algorithms effectively target the same modal peaks without producing any abrupt dips. By contrast, the full-range Wiim RC (20–20 kHz) shows more variability above 300 Hz, which highlights the importance of limiting correction to where it’s really needed—primarily below about 300 Hz in most rooms.

Another interesting observation is how the uncorrected response (green) may already be fairly balanced in parts of the midrange and treble, indicating that the LS60’s native design is solid beyond the low-frequency room interactions. The upshot is that if your speaker is already well-engineered (like the LS60), you usually won’t need significant mid or high-frequency adjustment, reinforcing the idea that limiting the correction range can yield the best overall result.
 
Sorry, I haven't read all the replies on this so maybe it's been asked already...

What's the consensus: bad WiiM correction algorithm or bad measurement or both?

If it's a bad algo then using a calibrated measurement mic might not help much.
I just posted a follow-up, and from the new measurements it’s clear the WiiM algorithm can be highly effective. The reason for the excessive boost below 100 Hz in previous tests was almost certainly the iPhone mic’s natural low-frequency roll-off, which WiiM most likely don't currently account for. Once they implement support for calibration files, it should become a remarkably capable and cost-effective room correction solution—almost too good to be true at the WiiM Ultra’s price point.
 
Follow-Up with Umik-1 Measurements

I revisited the Wiim RC measurements using a iPad Mini (borrowed from a friend) in combination with an Umik-1 microphone The measurement procedure remained the same as before, relying on the REW MMM (Moving Microphone Method) to capture averaged in-room responses.
  • Purple = Wiim RC (20–300 Hz)
  • Orange = REW MMM (20–300 Hz)
  • Red = Harman target curveView attachment 420288

From Figure 1, it is clear that Wiim RC (20–300 Hz) (purple) closely aligns with REW MMM (20–300 Hz) (orange) when using the Umik-1. This suggests that the Wiim RC algorithm is highly effective for low-frequency correction without calibration file.

By contrast, in Figure 2, the Wiim RC (20–20 kHz) (blue) introduces significant discrepancies at higher frequencies—most likely because the Wiim software cannot apply a calibration file for the Umik-1, leading to less accurate results above the midrange.
  • Green = Uncorrected
  • Purple = Wiim RC (20–300 Hz)
  • Blue = Wiim RC (20–20 kHz)
  • Orange = REW MMM (20–300 Hz)
  • Red = Harman target curve
View attachment 420289

Conclusion: When Wiim RC is limited to 20–300 Hz and measured with a Umik-1, the outcome is very similar to REW MMM. This indicates Wiim’s room correction can be excellent, as long as the software has accurate measurement data and is restricted to the problem frequencies below ~300 Hz. When Wiim adds support for custom calibration file, the RC solution could become even more robust across the entire bandwidth.

One additional point worth noting is how consistently both Wiim RC (20–300 Hz) and REW MMM (20–300 Hz)maintain a smoother transition between sub-bass and mid-bass. If you look at the 70–150 Hz region, both curves follow a very similar contour, suggesting the algorithms effectively target the same modal peaks without producing any abrupt dips. By contrast, the full-range Wiim RC (20–20 kHz) shows more variability above 300 Hz, which highlights the importance of limiting correction to where it’s really needed—primarily below about 300 Hz in most rooms.

Another interesting observation is how the uncorrected response (green) may already be fairly balanced in parts of the midrange and treble, indicating that the LS60’s native design is solid beyond the low-frequency room interactions. The upshot is that if your speaker is already well-engineered (like the LS60), you usually won’t need significant mid or high-frequency adjustment, reinforcing the idea that limiting the correction range can yield the best overall result.
Wiim are going to be adding an option to use multiple measurements in the near future, which should allow for even better results. I'm unsure whether it will follow the REW MMM approach or more akin to Audyssey with fixed mic positions.
 
I just posted a follow-up, and from the new measurements it’s clear the WiiM algorithm can be highly effective. The reason for the excessive boost below 100 Hz in previous tests was almost certainly the iPhone mic’s natural low-frequency roll-off, which WiiM most likely don't currently account for. Once they implement support for calibration files, it should become a remarkably capable and cost-effective room correction solution—almost too good to be true at the WiiM Ultra’s price point.
MiniDSP should put a switch on the UMIK that turns on a USB storage emulation mode so you can copy a calibration file to it. And if the file has the magic name then when you reboot the mic with that switch turned off, it applies the calibration. Now that's what I would call mini DSP.
 
MiniDSP should put a switch on the UMIK that turns on a USB storage emulation mode so you can copy a calibration file to it. And if the file has the magic name then when you reboot the mic with that switch turned off, it applies the calibration. Now that's what I would call mini DSP.
Neat idea, though would easily 1.5-2x the UMIK's MSRP.
 
OTOH, that phone is better equipped to handle the compensation,
Aside from the better equipment, it's actually pretty silly to do the compensation on every PCM sample in the mic. All you need is to compensate the measured spectrum that will be used, some dozens of computations as opposed to 48 thousand every second.

which developers like @Greg Wilding are already making use of.
HouseCurve. Nice! SO has an iPhone so I may try it out.

Hopefully WiiM will follow suit.
Yes.
 
Last edited:
Back
Top Bottom