Windows Setting Changes Frequency Response

[Presently42]

I first noticed something was amiss, when I measured my new JBL A130, and saw a bizarrely similar frequency response (fr) to what my old Technics SB-P1000 were producing. The measurements also didn't correlate nicely with what'd been measured here. Fast forward a while and much exasperaish later, and I discovered the culprit: enable audio enhancements (eae), supposedly required to use Equaliser APO was changing the sound in several ways

1. eae adds a hump around 10 kHz and 5 kHz; as well as something under Schroeder
2. eae crossfeeds, sending some of the left channel in to the right; and vice-versa
3. eae seems to put a shelving filter around 700 Hz

I've contacted the maker of Equaliser APO: since it works perfectly well under Windows 10 and Realtek drivers using SFX and no eae, there's no need to recommend it - let alone put the setting back on at every reboot. However, we'd like to know exactly what Windows is doing with this eae, and what its justificaish is. Furthermore, I've seen at least on fr chart here, which exhibited the eae humps: so this post is a bit of a psa too.

Anyone have a clue? Are there any current or former Microsoft employees here? I vaguely recall reading of an AES presentaish given by some Microsoft employees years ago, concerning audio enhancements - but I know not more.

Here are some measurements, with a legend:

 

[amirm]

@j_j and team designed the audio stack with that pipeline in it. Post all of our departures though, I heard it is not maintained or is broken. JJ may know better.

 

[Presently42]

Hmmm. Have you heard of the changes I've measured? If not, this might be a Realtek issue, rather than a Windows one

 

[amirm]

No, I don't use your setup.

 

[Presently42]

Hmmm, so nothing going on through the grapevine, then. I wonder if there are other Windows users on here, who have both Realtek (preferably) and maybe Equaliser APO, who might be willing to confirm my measurements

 

[PabloT]

Have you tried disabling pre-installed APOs?
Maybe following this guide could help:
https://www.audiosciencereview.com/...nding-the-windows-audio-quality-debate.19438/

 

[j_j]

My understanding is that the default for the resampler was set to "fastest". At one point for W7/W10 there was a way to change that.

It's been a long time, and I've long since fled that scene.

 

[Presently42]

Have you tried disabling pre-installed APOs?
Maybe following this guide could help:
https://www.audiosciencereview.com/...nding-the-windows-audio-quality-debate.19438/

This seems to be impossible for my setup:

Note, that headphone and speaker don't produce the same sound either: speaker drastically alters the fr, whereas headphone seems (I hope!) to not alter the fr. The first item is the hdmi out, which is just for the telly: my avr is too old!

 

[AluminiumEar]

Can confirm, my z370 rmaa test

 

[Presently42]

That's quite interesting. The curve you're showing isn't quite the one I've been experiencing. However, this does further confirm my suspicions, that the boosts being made by eae are related to Blauert bands. Interesting!

Anyway, your table doesn't note the massive crosstalk that eae also makes happen.... Can you tell us exactly your setup?

 

[j_j]

By Blauert's Bands, do you mean ERB's? https://en.wikipedia.org/wiki/Equivalent_rectangular_bandwidth

 

[McFly]

Not sure if entirely relevant but; recently I got a new Dell win10 laptop that's been hopeless for audio. You'll have to forgive my lack of computer knowledge but-

It has some app that runs in silent parallel with the realtek audio drivers called Maxx Audio Pro - this was adding all sorts of crap to the audio pipeline ridiculous bass boosts, cross-feeds and 3d hall effects to anything that was plugged in. With it uninstalled the headphone jack on the laptop would just full stop NOT EVEN WORK. No output.

Second to this there is some garbage Intel Audio USB controller that is also muddying the waters that can only be disabled not uninstalled (reinstalls if laptop restarted). This Intel controller seems to overrule any available audio sample rates a USB device reports and limits you to 16 bit 32khz max on USB mics and USB audio inputs. This controller had to be disabled for me to even use a Dayton DATS, to get the sound device reporting the controller as a "Generic USB audio device" and not "Intel Smart USB Audio Controller" . Hmm not so damn smart.

I bought the laptop PRIMARILY for audio use, input and output, and so far its been nothing but a pain in the ass.

 

[Offler]

I first noticed something was amiss, when I measured my new JBL A130, and saw a bizarrely similar frequency response (fr) to what my old Technics SB-P1000 were producing. The measurements also didn't correlate nicely with what'd been measured here. Fast forward a while and much exasperaish later, and I discovered the culprit: enable audio enhancements (eae), supposedly required to use Equaliser APO was changing the sound in several ways

1. eae adds a hump around 10 kHz and 5 kHz; as well as something under Schroeder
2. eae crossfeeds, sending some of the left channel in to the right; and vice-versa
3. eae seems to put a shelving filter around 700 Hz

I've contacted the maker of Equaliser APO: since it works perfectly well under Windows 10 and Realtek drivers using SFX and no eae, there's no need to recommend it - let alone put the setting back on at every reboot. However, we'd like to know exactly what Windows is doing with this eae, and what its justificaish is. Furthermore, I've seen at least on fr chart here, which exhibited the eae humps: so this post is a bit of a psa too.

Anyone have a clue? Are there any current or former Microsoft employees here? I vaguely recall reading of an AES presentaish given by some Microsoft employees years ago, concerning audio enhancements - but I know not more.

Here are some measurements, with a legend:

View attachment 140942
View attachment 140943

I dont know if the issue can be related to this:
https://www.audiosciencereview.com/...indows-audio-quality-debate.19438/post-807219

In short - I made EQ APO work on my Realtek soundcard (Troubleshooting options, install as SFX/EFX), endpoint is Optical SPDIF to external DAC of D3020v2.

If Preamp is set to -0,14dB or lower there is no added distortion (aside from specs of amp). If its set to anything higher than -0,14dB it introduces distortion as audiolimiter kicks in. I measured that on both FFT graph and frequency sweep.

In the test there are no other changes made, just preamp in system is decreased.

Root cause issue is that audiolimiter (often referred as CAudioLimiter) kicks in and does not allow volume beyond certain threshold, effectivelly distorting the sinewave. That can be observed also on oscilloscope.

 

[restorer-john]

Throw it all out and go back to CDs.

 

[Grotti]

Throw it all out and go back to CDs.

A Streamer and a HDD will do the job too. Use your Computer as a remote control

 

[Presently42]

By Blauert's Bands, do you mean ERB's? https://en.wikipedia.org/wiki/Equivalent_rectangular_bandwidth

Indeed not: instead, I was noting the work of Professor Blauert, a German doctor who discovered frequency bands responsible for various imaging effects. I'm afraid I've been heretofore unable to find his own published work: possibly I'd have to search in German. In any event, a notable excerpt of his work is on the German Wikipedia page, a terrible googly translate English version of which can be found here

 

[j_j]

Indeed not: instead, I was noting the work of Professor Blauert, a German doctor who discovered frequency bands responsible for various imaging effects. I'm afraid I've been heretofore unable to find his own published work: possibly I'd have to search in German. In any event, a notable excerpt of his work is on the German Wikipedia page, a terrible googly translate English version of which can be found here

I suspect you will find a strong equivalence.

 

[Presently42]

I suspect you will find a strong equivalence.

Hmmm, from my understanding of ERBs, they discuss the ear's ability to perceive differences in frequencies: the higher one goes, the greater the difference between two frequencies before a difference is heard - hence why small filters at high frequencies aren't very useful. The Blauert bands discuss how specific bands of frequencies are at least partially responsible for localisaish: see this post for a rather confusing description of the bands; or this image for a clearer graph. Note, that @Amir has discovered a rather similar phenomenon with headphones: boosting the 2.7 kHz to 5.5 kHz region increases the stereo effect. I've independently also found, that increasing the region close to 8 kHz has a similar effect, which professor Blauert describes as height. I believe, that Dr. Toole has come to the same conclusion.

I suspect, that what I was measuring in my original post, was Realtek (for I now believe Realtek to be the culprit, and not Windows) trying to emulate what the gentleman in the post I linked above was trying to achieve; namely, small changes (eq boosts) to the fr, which coincide with Blauert bands, thereby enhancing the stereo image. Unfortunately, such enhancements - or at least, their interpretaish - aren't terribly pleasant or efficacious.

 

[Guermantes]

Not sure if entirely relevant but; recently I got a new Dell win10 laptop that's been hopeless for audio. You'll have to forgive my lack of computer knowledge but-

It has some app that runs in silent parallel with the realtek audio drivers called Maxx Audio Pro - this was adding all sorts of crap to the audio pipeline ridiculous bass boosts, cross-feeds and 3d hall effects to anything that was plugged in. With it uninstalled the headphone jack on the laptop would just full stop NOT EVEN WORK. No output.

Second to this there is some garbage Intel Audio USB controller that is also muddying the waters that can only be disabled not uninstalled (reinstalls if laptop restarted). This Intel controller seems to overrule any available audio sample rates a USB device reports and limits you to 16 bit 32khz max on USB mics and USB audio inputs. This controller had to be disabled for me to even use a Dayton DATS, to get the sound device reporting the controller as a "Generic USB audio device" and not "Intel Smart USB Audio Controller" . Hmm not so damn smart.

I bought the laptop PRIMARILY for audio use, input and output, and so far its been nothing but a pain in the ass.

I've been struggling with a new Dell laptop (Inspiron 5510), too. Using it with a MOTU Ultralite Mk4 and periodically having audio glitches even with ASIO drivers. The culprits for me are ACPI.sys (power management) and storport.sys (storage access) processes causing large DPC latency spikes. The former is apparently a common problem with Windows laptops, the latter is possibly related to my SSD.

Oh, and I found shifting the audio interface over to the USB-C/Thunderbolt port improved other USB-related issues.

 

[digitalfrost]

storport.sys (storage access) processes causing large DPC latency spikes. The former is apparently a common problem with Windows laptops, the latter is possibly related to my SSD.

You could try to disable C6 sleep, but it will impact battery life. Have you tried if changing to the High Performance Power Plans changes anything?