Room correction: very strange Per Channel algorithm and it's respective results

[AlxTMX]

Hi there!
While I'm very happy with Stereo Room Correction after I has implement an appropriate settings for Frequency, Max Gain and Max Q, the Per Channel Room Correction is a big mistery for me.
In theory the aim should be to lead both channels that could sounds differently to the same target curve. So one should feed the very same test signal to the left and right channel and compensate their deviations appropriately to make them sounds the same.
In real world Room Calibration Algoritm utters the full spectrum 20-20K sweep tone on the Left and then some modulated noise fragment on the Right. As a tresult the correction for Left channel is very same to Stereo correction curve. But in the noise fragment for th Right channel is fery few of low frequencies, it's in mid range and highs at most. So the Right channel correction curve is become huge overcompensated in bass region and depressed in higs.
The multipass Per Channel mode is almost same: the first test fragment for the Left is 20-20K sweep tone is followed by two modulated noise fragments but then there is only three modulated noise fragments for the Right. And the results are the same as in one pass mode due to the Right channel test tones are lack of bass and the coreection curve is ovecompensate this imaginary problem.
All of it is very strange for me. What's the aim of this assimetrical algorithm?
I'm using an external USB soundcard and an electret measuring microfone but that's not the point. One can hear by ears that the test signals are completely different so the results are inappropriate.

 

[slartibartfast]

Hi there!
While I'm very happy with Stereo Room Correction after I has implement an appropriate settings for Frequency, Max Gain and Max Q, the Per Channel Room Correction is a big mistery for me.
In theory the aim should be to lead both channels that could sounds differently to the same target curve. So one should feed the very same test signal to the left and right channel and compensate their deviations appropriately to make them sounds the same.
In real world Room Calibration Algoritm utters the full spectrum 20-20K sweep tone on the Left and then some modulated noise fragment on the Right. As a tresult the correction for Left channel is very same to Stereo correction curve. But in the noise fragment for th Right channel is fery few of low frequencies, it's in mid range and highs at most. So the Right channel correction curve is become huge overcompensated in bass region and depressed in higs.
The multipass Per Channel mode is almost: the first test fragment for the Left is 20-20K sweep tone is followed by two modulated noise fragments but then there is only three modulated noise fragments for the Right. And the results are the same as in one pass mode due to the Right channel test tones are lack of bass and the coreection curve is ovecompensate this imaginary problem.
All of it is very strange for me. What's the aim of this assimetrical algorithm?
I'm using an external USB soundcard and an electret measuring microfone but that's not the point. One can hear by ears that the test signals are completely different so the results are inappropriate.

The left and right test signals should be the same so report the issue to WiiM via feedback in the app.

 

[Wiimer]

Hi there!
While I'm very happy with Stereo Room Correction after I has implement an appropriate settings for Frequency, Max Gain and Max Q, the Per Channel Room Correction is a big mistery for me.
In theory the aim should be to lead both channels that could sounds differently to the same target curve. So one should feed the very same test signal to the left and right channel and compensate their deviations appropriately to make them sounds the same.
In real world Room Calibration Algoritm utters the full spectrum 20-20K sweep tone on the Left and then some modulated noise fragment on the Right. As a tresult the correction for Left channel is very same to Stereo correction curve. But in the noise fragment for th Right channel is fery few of low frequencies, it's in mid range and highs at most. So the Right channel correction curve is become huge overcompensated in bass region and depressed in higs.
The multipass Per Channel mode is almost same: the first test fragment for the Left is 20-20K sweep tone is followed by two modulated noise fragments but then there is only three modulated noise fragments for the Right. And the results are the same as in one pass mode due to the Right channel test tones are lack of bass and the coreection curve is ovecompensate this imaginary problem.
All of it is very strange for me. What's the aim of this assimetrical algorithm?
I'm using an external USB soundcard and an electret measuring microfone but that's not the point. One can hear by ears that the test signals are completely different so the results are inappropriate.

When the L/R RC was tested, there were no problems with the test sweep and the left and right measurements were exactly the same.

Have you tried rebooting the device and phone?

 

[AlxTMX]

Have you tried rebooting the device and phone?

Huh! You're lucky man! Sure I've rebooted my devices as it's the very first option for me as I'm an IT professional with thirty years of expirience
Note this assimetrical signals are typical for my both WiiM Pro Plus and just bought Ultra devices. Most corrent FW and the app too.

 

[slartibartfast]

Huh! You're lucky man! Sure I've rebooted my devices as it's the very first option for me as I'm an IT professional with thirty years of expirience
Note this assimetrical signals are typical for my both WiiM Pro Plus and just bought Ultra devices. Most corrent FW and the app too.

Android or iOS? Maybe reinstall the app. I'll try later but it has always worked for me previously.

 

[canard]

When the L/R RC was tested, there were no problems with the test sweep and the left and right measurements were exactly the same.

Have you tried rebooting the device and phone?

View attachment 15454

This perfect correspondence between the l and r results is more than suspect... do others observe it too?

 

[Wiimer]

This perfect correspondence between the l and r results is more than suspect... do others observe it too?

No, nothing suspicious.

Both were tested with a test sweep output from the left speaker. That is, the "L" plug stuck in the Ultra's RCA output jack was connected to the "R" jack before the second "R" measurement was taken.

 

[canard]

No, nothing suspicious.

Both were tested with a test sweep output from the left speaker. That is, the "L" plug stuck in the Ultra's RCA output jack was connected to the "R" jack before the second "R" measurement was taken.

but the results have little reason to be equally comparable on the two channels...it seems to me...very improbable...

 

[Wiimer]

but the results have little reason to be equally comparable on the two channels...it seems to me...very improbable...

This is a simple test to see if the L and R sweep signals match.

 

[canard]

sorry ...
misunderstood...

It will be good if wiim suggests a multi-measure approach, methodology...what they have integrated as an approach...

 

[AlxTMX]

Android or iOS? Maybe reinstall the app. I'll try later but it has always worked for me previously.

Thanks! I can try. I don't know if this will help, but it's not hard to do.
In general, I was sure that the entire algorithm is in the cloud and the application is only for display, since Room Correction does not work when the Internet is disconnected.

 

[canard]

Thanks! I can try. I don't know if this will help, but it's not hard to do.
In general, I was sure that the entire algorithm is in the cloud and the application is only for display, since Room Correction does not work when the Internet is disconnected.

I don't really understand why in fact...the computing power of our phones and tablets is so high that it would be a joke for them ;-)

 

[hgo58]

I don't really understand why in fact...the computing power of our phones and tablets is so high that it would be a joke for them ;-)

Probably to secure the algorithms used?

 

[AlxTMX]

I don't really understand why in fact...the computing power of our phones and tablets is so high that it would be a joke ;-)

Cloud computing have it's benefits and negatives. Here the benefit is that programmers can modify algorithm and test signals without forcing users to update their apps. But one can see that things are not so simple in real.

 

[AlxTMX]

Android or iOS? Maybe reinstall the app. I'll try later but it has always worked for me previously.

No, it doesn't work for me unfortunately. I have make factory reset for my device and reinstalled the app but with no luck. I've got the same different test signals on Left and Right as previously.

 

[slartibartfast]

No, it doesn't work for me unfortunately. I have make factory reset for my device and reinstalled the app but with no luck. I've got the same different test signals on Left and Right as previously.

Have you sent feedback to WiiM?

 

[AlxTMX]

Have you sent feedback to WiiM?

No. I've cought the problem. I'll wright about it tomorrow but there is something to think about for WiiM team.

 

[harkpabst]

No, it doesn't work for me unfortunately. I have make factory reset for my device and reinstalled the app but with no luck. I've got the same different test signals on Left and Right as previously.

Did you try reversing the left and right output of then WiiM Pro Plus or WiiM Ultra? Or reversing the cables to the left and right speaker?

The sweep signal should be the same for both channels.

 

[AlxTMX]

Hi, colleagues! So, I've caught the problem as I said previously.
I'm used to use WiiM as streaming digital transport connected to classy external DAC using Coaxial or Optical interface. The interface doesn't matters in this particular case 'cause both support up to 24/192 and worked flawlessly for more than a year (for Pro Plus) in various streaming scenarios (Tidal, Qobuz etc). DPLL set to minimum/off and no problem at all.
The problem has appeared when Per Channel Room Correction has been developed byWiiM. The first fragment (applause) is in 16/44.1KHz. Then folows the Left channel sweep tone in 48KHz. And at the very end of Left sweep tone I hear a noise pattern for a fraction of second. All next sweeps - Right channel sweep or two Left channel and three Right channel sweeps in multipass scenario - are sounds as modulated noises for me.
There is nosuch problem on analogue interfaces of WiiM devices.
The problem coluld be solved if one set any fixed resolution for digital interfaces or if set the resolution to 44.1 that is the same as it's the single minimum value.
The problem persists if I set 48KHz max resolution but not fixed. So I presume that the problem caused by dropping down sinchro to 44.1 at the end of left sweep fragment or something like that.
Thee problem doesn't exists in a case of Stereo sweep tone - neither in single nor in multipass scenario.
I suspect it's Left channel sweep tone guilt. Can't download to chec the file. Hope the WiiM team could double chec this Left channel sweep tone if it's w/o some strange artifacts at the very end?
Thanks in advance!

 

[Maurizio]

Hi there!
While I'm very happy with Stereo Room Correction after I has implement an appropriate settings for Frequency, Max Gain and Max Q, the Per Channel Room Correction is a big mistery for me.
In theory the aim should be to lead both channels that could sounds differently to the same target curve. So one should feed the very same test signal to the left and right channel and compensate their deviations appropriately to make them sounds the same.
In real world Room Calibration Algoritm utters the full spectrum 20-20K sweep tone on the Left and then some modulated noise fragment on the Right. As a tresult the correction for Left channel is very same to Stereo correction curve. But in the noise fragment for th Right channel is fery few of low frequencies, it's in mid range and highs at most. So the Right channel correction curve is become huge overcompensated in bass region and depressed in higs.
The multipass Per Channel mode is almost same: the first test fragment for the Left is 20-20K sweep tone is followed by two modulated noise fragments but then there is only three modulated noise fragments for the Right. And the results are the same as in one pass mode due to the Right channel test tones are lack of bass and the coreection curve is ovecompensate this imaginary problem.
All of it is very strange for me. What's the aim of this assimetrical algorithm?
I'm using an external USB soundcard and an electret measuring microfone but that's not the point. One can hear by ears that the test signals are completely different so the results are inappropriate.

Hi @AlxTMX
Unfortunately, you are not the only one who has experienced this defect.
I also had, and have, it from the first moment with the Ultra connected via USB to my Topping D70ProSabre DAC.
If I use ARC per channel, from L channel the sweep tone starts in the correct way, although it ends in a couple of seconds of “modulated noise,” as you called it. I called it white noise. When switching to the R channel, no sweep tone, only modulated noise or white noise.
Of course, the equalized curve of the R channel is completely meaningless. As you say "huge overcompensated in bass region and depressed in higs".
In my case, given the position of the speaker (rear bass reflex) which is placed in an angle I should have achieved exactly the opposite. Cut bass and increased treble. I tried everything. Rebooting the Ultra; reinstalling the App; I changed from IOS to Android, but the defect is still there.
On the other hand, if I use the ARC per channel with the DAC connected to the Ultra with a COAX cable, everything works at his best. The problem is only with the USB.
I opened two tickets through WHA attaching the log, but I got no response from @WiiM Support or @WiiM Team .
I thought I was the only one with that problem, but now I see that I am not alone. I hope that @WiiM Support or @WiiM Team can solve this annoying problem.