Audiolense 2.0 channel Roon filter sounding like echo between left and right channels

[jrobbins50]

Hello, crew. As our home remodeling continues, I finally get to put my secondary system on our porch back together. Let me start by saying that the filters I last created in June 2020 are working fine. But, we've gone to tile flooring and I have a different pair of front speakers, so I clearly needed to do a new AL measurement and filter creation. I'm using an Audioquest Cobalt Blue in this system as my DAC. Measurement, filter creation and input into Roon all very smooth -- Bernt, you've made some nice changes to the software over the past nine months.

This said, with the old filters, or none at all, my speakers are imaging well and the vocalist is directly in the middle where I'd expect. But, when I use the new Roon filters I've created today (and I've tried several times, with different measurements), the left channel sounds normal and the right channel is attenuated and muffled, and also not in sync with the left channel -- almost like an echo.

I don't know what I'm doing wrong here. I'm attaching the measurement file and the zipped filter set for anyone to provide some help on this one. Cheers. JCR

(not sure the measurement set uploads....)

 

[juicehifi]

According to the correction, the right speaker is more than 1 meter further away than the left. If that sounds wrong you need to remeasure.

 

[jrobbins50]

Bernt, I finally got the measurement to work and the system is sounding great. Thanks. JCR

 

[jrobbins50]

Ahem, Bernt, I was wrong. The measurement is still not right. Omid was kind enough to take a look at a new measurement for me over the weekend and he saw the same thing you did: a volume level difference and delay between the two front mains. He suggested that I try some different drivers for measurement. As I've not had to do that in the past, I'm not sure what I should use. I have been using ASIO4all for both the Audioquest Dragonfly Cobalt DAC and the UMIK-1 mic. I should mention that I tried a different DAC, just for measurement, but that came out with the same results. To be clear, the sub is closer to the left main than to the right main and I'm measuring as a 2-channel arrangement, with left main + sub and right main + sub. Files attached for your review. Thanks for your thoughts. JCR

 

[juicehifi]

The measurement do not look OK. It has an unusual high noise floor and the timing seems to be off. You should look for technical problems along the whole audio path.

Try to flip left and right speaker out of the dac or use the channel override in Audiolense. And see if the delay moves over to the other speaker. If it does, you likely have dropout problems with the stream.

Listen to the recorded sweep. It should sound pretty clean. And it should look like a clean sweep without any discontinuities. You can use Audacity to do both. The sweep is stored in "sweepmeasurement.wav". This file is overwritten on every take, btw.

Look for ways to increase buffer sizes. Set the asio4all buffer to max.

Also, try to measure without using Asio4All. Engage "separate play and recording devices" and try the different methods ... and see where that gets you.

 

[jrobbins50]

Bernt -- kindly let me know how this set looks now. Thanks. JCR

 

[juicehifi]

Before I examine the files, jrobbins, what are your recent resultat?

 

[jrobbins50]

Before I examine the files, jrobbins, what are your recent resultat?

Bernt, I'm not sure I understand what you are asking by "resultat". If you are asking whether the sound now appears to be balanced, yes, that is so. If you are asking for something else, then kindly explain further. Thanks. JCR

 

[juicehifi]

I tried to ask what you want me to look for in those files. If you got the centered result I bet they are OK.

"resultat" is Iphone's autocorrect feature. I have Norwegian and English engaged, and it is always the wrong language that is activated...

 

[jrobbins50]

I'd appreciate if you could look at the impulse response. I continue to have my own personal difficulties in looking at that and I am concerned that it isn't showing the nice curve that I've seen in the past on other systems in my home. Thanks. JCR

 

[juicehifi]

IR of your latest measurement:

IR from the stereo demo measurement:

You don't always get as clean an impulse as in the demo measurement, but it will typically be max one or two peaks in both direction before you start to see a decay pattern.

In other words, somethings seems to be wrong with your measurement.

The frequency response may be close to what it should be, but the timing is wrong. So I suspect glitches in the measurement stream ... playback, recording or both.

 

[jrobbins50]

Ok that is very helpful, Bernt. I did see this pattern you are showing me and thought it seemed off, and you are confirming. So odd that I’m having the measurement difficulties. I’ve tried ASIO, WASAPI and the two other Windows sound variants. I will keep at it.

As to the measurement, is OK dynamic range acceptable (which is all I have achieved) or do I really need Good or Excellent? And how would I best increase the dynamic range for measurement — increasing the volume level of the playback sweep, increasing the microphone sensitivity on the measurement page or something else?

Thanks. JCR

 

[juicehifi]

Yes, OK dynamic range is acceptable. Best of luck with your efforts!

 

[jrobbins50]

OK, Bernt. Well, the sun porch system at hand is at a pause, as the center channel amp went out. I'll return to that when it's back from the repair shop.

Meanwhile, I have turned to my den system. I replaced the two channel amp in there and have re-run measurements (a simple 2.1 system treated as left main + sub and right main + sub). The correction results look fabulous, but the sound is off here as well. Specifically, I've lost the pinpoint imaging that I previously had. (These are KEF LS50 speakers for the mains, so they should image well.) The image is spot on when I turn convolution off in HQP (for this system, I saved the 384K filter as mono and then have the Den HQP 20_384.wav file as the left channel in the HQP convolution matrix and the Den HQP 21_384.wav file as the right channel.

I use the Stereophile test CD1, with Sam Tellig having a dog bark, to test imaging. Normally, the in phase dog bark is centered between the two mains and the out of phase dog bark is diffuse. That's what I get when I turn off the HQP convolution and thus these new AL filters. But, with the convolution enabled, the dog bark is diffuse and not centered either in or out of phase.

I measured with the auto polarity enabled, but I assume that if it inverted incorrectly, the test track would center the dog bark on the out of phase recording. Of course as noted above, it doesn't.

I'm attaching the measurement file and the 384 mono filter set and corresponding wav files and ask if you can take a look for me and advise what I might do differently. One thing is for sure: the measurement shows a crazy 27 dB swing between the peak at around 60 Hz to -22 dB at 200 Hz, and the AL software with 8.00 dB maximum correction boost totally flattens everything out. It looks very impressive -- I just wish it sounded as good! Thanks. JCR

I appreciate your help. Cheers. JCR

 

[jrobbins50]

Bernt, bumping this one back up to the top for your consideration. Thanks. JCR

 

[juicehifi]

Hi JRobbins,

SHort version: I don't think the measurement is OK.

In other words: Same problem as before.

Are you able to make a measurement with a proper impulse response in REW?

 

[jrobbins50]

Apparently, I am not able to get a good impulse response measurement, Bernt. I can get measurements recorded with ASIO4all for the mic and ASIO4all, WASAPI, Direct Sound and MME for the playback device (my LHLabs DAC). I’ve also tried with and without enabling polarity correction. Each measurement is essentially the same — I can use the measurement to generate a correction file that very much matches me to the frequency target, but the impulse response continues with the wild pre- and post-ringing you’ve seen. I had this system working great with AL last fall before we remodeled, so I am sure that it should be doable.

At this juncture, I am wondering if the measurement sweep file is somehow corrupted (and where might I find a replacement for it to check) or if my UMIK-1 has gone bad (in which case, I could order a UMIK-2 at this point).

Your further thoughts are appreciated. Cheers. JCR

 

[juicehifi]

I think you need to device a proces where you can isolate the problem by elimination. E.g. try another microphone, another sound card, another PC etc.

 

[jrobbins50]

I’ve ordered up a UMIK-2 mic, which will take a few weeks to arrive from miniDSP. I have another 2-channel soundcard that I will try. I’ve been doing measurements on my laptop, but I do have another PC in my den that I could install AL onto and try it for measuring. Remind me, Bernt, what I need to do for an AL license for this additional PC. Finally, is there a place to download the measurement sweep file? I’d like to get a fresh version to ensure mine is not somehow corrupted. Thanks. JCR

 

[juicehifi]

You can just send me a hardware id for another machine. No problem!

 

[jrobbins50]

You can just send me a hardware id for another machine. No problem!

Bernt, I sent a private message to you with the hardware ID info. Thank you. JCR

 

[juicehifi]

Mail?

 

[jrobbins50]

Hi Bernt. I sent the hardware IDs (from the AL software) to you at your juicehifi email address. Thanks. JCR

 

[jrobbins50]

Bernt, an update. Instead of using my laptop for running AL and recording measurements, I used the den PC that runs Roon bridge (an old i-3 processor) and experimented among ASIO and the various Windows choices. I forgot to document which choice worked, but the second set of filters had pinpoint imaging of the dog bark when the Stereophile test track was out of phase. When it was in phase, the dog bark was diffuse.

I physically swapped the leads to my right front speaker and voila, sounding very nice. I did have auto polarity turned off, but with my laptop, it made no difference whether it was on or off. However, it’s clear with the filters generated by the PC that the polarity was wrong in my system. In MCH mode in this same system, I detect no sonic difference from having physically changed the speaker polarity. So, this seems to be a win.

Meanwhile, my new UMIK-2 mic arrived. I note that it can do 192K recording. Of course, 48K is the UMIK-1 standard. Is there any reason for me to try the new mic at a higher rate and if so, what do you think would be the right choice? Thanks. JCR

 

[juicehifi]

Imo no need to measure at high sample rate