sprite
Post image

Xonar drivers disable SpeakerCompensation tweak testing

What it does (updated 03 Aug 2017)

This tweak disables the SpeakerCompensation driver setting, this setting is enabled by default and not accessible through Asus/C-media Panel. The purpose of SpeakerCompensation was unknown but from what I was able to deduce, when enabled, SpeakerCompensation is altering the sound output by modifying the amplitude of sound frequencies. By disabling it, early indications show that the sound output may be closer to the source file. Disabling SpeakerCompensation also results in a lower DPC latency. There may be other benefits or drawbacks that are yet to be discovered.

I invite you to test this tweak and submit your findings and impressions. I’m interested in hearing if you find the sound quality to be better after disabling “SpeakerCompensation” and if you consider that this tweak should be applied by default with UNi Xonar drivers.

Tweak works for the all the UNi Xonar supported cards:

  • Asus Xonar: DG, DGX, DG SI, DS, DSX, D1, D2, DX, D2X, HDAV, HDAV Slim, ST, STX, STX II, Xense.
  • Other C-Media CMI8788 audio chip based cards: Auzentech X-Meridian & X-Meridian 2G; HT Omega Claro Plus, 2, Halo, eClaro; Razer Barracuda AC-1.

Page updates

  • 16 Oct 2017: Updated “Audio frequency analysis”. Added “Possible performance problems” section.

How to apply the tweak

For easy switching between enabled and disabled, it’s recommended that you use UNi Xonar v1.80 or older drivers.

  • Download and extract this file.
  • Run “disable SpeakerCompensation.exe”.
  • For v1.80 and older drivers: Close any application that uses the soundcard and run “restart card.exe”. You would hear the card popping if the restart went through, if you didn’t hear that then restart the PC.
  • For v1.81 drivers: Restart the PC. Do not run “restart card.exe” as it will result in a blue screen.
How to restore to the default behavior
  • Run “restore SpeakerCompensation.exe”.
  • For v1.80 and older drivers: Close any application that uses the soundcard and run “restart card.exe”. You would hear the card popping if the restart went through, if you didn’t hear that then restart the PC.
  • For v1.81 drivers: Restart the PC. Do not run “restart card.exe” as it will result in a blue screen.

Audio frequency analysis (updated 16 Oct 2017)

On my Xonar DX card using a short loop cable I’ve recorded the playback of various melodies with SpeakerCompensation enabled and disabled. For each recording I’ve generated a FFT analysis spectrum and compared it to those of the original files. I’ve observed that when SpeakerCompensation is enabled it significantly decreases the amplitude of the frequencies that are above 21200 Hz and slightly alters other frequencies.

For an easier comparison I’ve made a audio file with the right channel from original file and the right channel from each recording. I’ve aligned each track(channel) of the resulted files so that they where exactly the same and then generated the FFT analysis spectrum of a specific segment. You can check some of the resulted FFT analysis spectrums done with Adobe Audition below. Images will be resized for screens below 1920x1080px so you might want to save and view them locally.

Beside the obvious amplitude decrease of the frequencies that are above 21200 Hz, when comparing the FFT spectrums it looks like sometimes the recording with SpeakerCompensation disabled is more in line with the original file, and sometimes the recording with SpeakerCompensation enabled seem to be.

Some of these difference could be recording artifacts. These recordings where restricted to my Xonar DX card’s playback and recording capabilities so better audio cards could produce more accurate recordings. Different audio tracks or segments could show different results. I may do other FFT spectrums of different audio files later, the process of doing these FFT spectrums is very time consuming.

On the topic of of the frequencies above 21200 Hz (21.2 kHz), I’ve read that commonly stated range of human hearing is 20 Hz to 20 kHz, but not all of us can hear up to 20 kHz and that some can hear above 20 kHz . So if frequencies above 20 kHz can’t be heard then why bother decreasing them? Is there any benefit of doing this? Maybe someone from the community can help here.

Audio quality (Work in progress)

The majority of the people that provided feedback so far, reported that the tweak improves the sound quality with the sound being richer, more clearer and detailed. Not everyone will notice a improvement in sound quality. The combination of ones hearing capabilities (check these tests), type of music they listen to and speakers/headphones performance are the determining factors in whether you can hear an improvement with this tweak or not.

Other people’s input is welcomed. Please check the feedback guidelines.

Right Mark Audio Analyzer (RMAA) tests

RMAA tests that I’ve done on my Xonar DX with a audio loop cable didn’t show any consistent differences between “SpeakerCompensation” being disabled and enabled.

Possible performance problems

There is a small possibility that on some systems the performance is affected by applying this tweak. Only 2 people reported such behavior.

I’ve ran multiple game tests on my PC and there where no FPS difference, no FPS drops or freezes between having SpeakerCompensation disabled and enabled. On Windows 7 with UNi Xonar v1.80a I’ve tested Dota 2 and Mafia 2, on Windows 10 with UNi Xonar v1.80b I’ve tested Hitman 2016 (DX11 and DX12) and Prey 2017.

At the moment we don’t know what causes the performance issues, but probably these are caused either by some specific audio settings or in conjunction with specific hardware. Anyone having performance problems when SpeakerCompensation is disabled please tell me the following information which audio driver version you have installed, panel configuration, number of speakers/channels, analog or digital(S/PDIF) output, card sample rate, CPU, videocard, OS.

DPC Latency results

DPC latency tests showed a 10-25% lower DPC latency that’s been generated by the use of the Xonar sound card when SpeakerCompensation was disabled. Results vary depending on other soundcard settings that are used and OS being tested.  Some tests I ran:

  • Windows 7, measured with DPC Latency Checker, network disabled, UNi Xonar v1.80b Low DPC Latency, Aimp audio player, avg. DPC latency results – DPC Latency in idle(10us):
speakercompensation2.0 analog5.1 analog7.1 analog7.1 analog with ASUSAudioCenter
disabled203641115
enabled254557134

 

  • Windows 10, measured with LatencyMon, UNi Xonar v1.80b Low DPC Latency, 2.0 analog output, Aimp audio player, 2 min audio playing, results for STXII.sys:
speakercompensationDPC countHighest execution(ms)Total execution(ms)
disabled599970.0660981034
enabled650870.0737671154

Feedback guidelines

When submitting feedback please include the following: audio card, audio driver version, Windows version, settings used (analog or digital output, number of channels, any other relevant settings that where enabled), speakers and amp or headphones model, music format (lossy or lossless).

With “SpeakerCompensation” disabled I am also interested in knowing up to which frequency you can hear (see tests here).

 

Advertisements

136 Comments

  1. Sranken Ftein27 Sep 2017 @ 22:17

    Hello! Thanks for caring about this driver so much, amazing stuff. To the topic:

    I listened to mutliple Songs, with different instruments and styles, listening them back and forth, some directly from the cd, some flac and some mp3. I feel like this change is either not doing much or it is a placebo (or not working for me somehow ofc). Every time I change to the “better” setting, it feels kind of better, but when I change back it is lacking nothing as far as I can tell compared to the presumably better version, which indicates a plecebo effect in my case. This does not mean this change is not working and everyone here is mad or something, (I actually think that that many people cant be so wrong) it is just my experience and maybe it did just not function properly. I kinda cannot blame it on my equipment either, especially because people here are reporting such a tremendous change, not a subtle one. A Beyerdynamic T90 should be enough to hear some difference or? Another way to explain at least some of it is, that it really only influences the very high and low spectrums, especially the low ones were not present in my testsongs. I did not test speaker yet. And yes, the restartsoundcard.exe made the card click every time. Also I let another person test it on one song, without knowing which should be better, she actually liked the “worse” one more. And I might add that a recent hearing test at my otologist attested me very good ears.

    Some Info:
    I use a STX II with the v1.80b r3 on win7. Like I said I used FLAC, Mp3 and direct CD playback. Using a Beyerdynamic T90. 44.1Khz. 24 bit 44.1Khz in win7 settings. Tried it with Dolby Headphone and wihtout. I will build my new pc soon with win 10 probably, I will report if it is a different experience. Which diver do you personally recommend for win10 stx II btw? Newest or 1.80b r3?

    Reply

    • CarvedInside30 Sep 2017 @ 02:13

      Hello. I understand what you are saying. To me the changes aren’t that obvious, if I listen closely I notice slight changes but not enough to say it’s obviously better, it might just be placebo. My hearing caps at 14kHz so that might be the reason why it isn’t obvious.
      While there might be some placebo involved for some people, it’s a measurable fact that SpeakerCompensation alters frequencies above 21kHz. In my FFT analyses I’ve seen alterations in the 15kHz range and above, but those alterations are not that drastic and might be a result of bugs within the program I use. I am having a hard time producing consistent results that I can be certain are correct, so I am still working on this part.
      The tweak most likely applied correctly to your system, but as you said the music you are listening may not have very high frequency spectrums. You can check a song’s freqeuncy spectrum with a tool like Ocenaudio(Analyze ->FFT analysis) or Audacity.
      Another factor the frequency up to which you can hear. Do these audio frequency tests with SpeakerCompensation disabled and let me know of the results.
      I don’t have a strong opinion on which driver to use, maybe install 1.80b and if you have any problems try the 1.81a.
      Thank you for the feedback.

      Reply

      • Sranken Ftein30 Sep 2017 @ 13:14

        Meanwhile I did the test again on my new rig, same experience. According to those samples I can hear up to 18kHz, so if its above 21kHz, that would explain things, but not if its above 15kHz. Btw: the 1.80b driver is running stable atm.

        I will do the frequency tests on the songs soon.

        Reply

    • Evgenim30 Sep 2017 @ 02:26

      No placebo at all. Me and my neighbour often listening to collection of trance music. The first that has been noticed it’s more clear hearable of high freqs. Low freqs also have some more detail. Listened to the same tracks all the time and we know how it’s sounds before tweak. Regards.

      Reply

  2. Dima29 Sep 2017 @ 18:07

    Windows 10, Driver 1.80 r3, Xonar DG pci-e, 2.0 speakers, 2 channel configuration, 44.1kHz, no SVN used.
    I can say with confidence this tweak allows me to push my speakers volume higher and get a cleaner sound than before. The highs and stereo sounds much cleaner.

    Reply

    • CarvedInside30 Sep 2017 @ 02:17

      Thank you for the feedback. Can you tell me up to which frequency you can hear in these tests?
      Just a clarification, when you change the volume you change the Windows volume or that on your amp/speakers?

      Reply

      • Dima30 Sep 2017 @ 20:23

        I can hear up to 22kHz. My windows volume is always at maximum. I turned my speakers up when testing this.

        Reply

        • Alex R19 Oct 2017 @ 00:32

          Placebo. Dima please explain how can you “push the volume higher” if FFT analysis shows no change in audio level whatsoever and only a minor change to inaudible high frequencies?

          Reply

  3. Asri10 Oct 2017 @ 18:34

    Win10’s antivirus claims, that the SpeakerCompensation tweak is a trojan. I don’t believe it is, but I wanted to inform You about this anyways. To be precise, Win10 antivirus claims its a “Win32/Tilken.B!cl” trojan. Maybe You could get in contact with the Microsoft guys and clear this up?

    Reply

  4. Carol Laszlo14 Oct 2017 @ 02:36

    Hy there boys, i have a great respect for your hard work as as an apreciacion i will donate.I have a question i am a Xonar D2X owner , running windows 10 PRO x64 with “creators update” installed. I can’t get my headphone to work with my card, it’s a simple jack headphone from phillips, using the backport (5.1 , green socket ) jacks. CAN SOMEBODY PLEASE GIVE ME THE LINK TO THE DRIVER I HAVE TO INSTALL TO MAKE IT WORK ? AS I DON’T UNDERSTAND ALLMOST ANYTHING FROM THEESE TECH STUFF, and i am getting really hopeless.

    Reply

  5. dramu14 Oct 2017 @ 22:24

    There are many speakers out in the wild that can fold ultrasound (non-audible frequencies) back into the audible region. Of course this is an unwanted phenomenon and it means if you send non-audible frequencies to your speakers, they might play them back as very much audible noise, depending on the electronics, mechanical speaker construction and the current sound material involved.

    What I’m saying is, this is most probably the reason why “speaker compensation” will attenuate from 21.2kHz. Those frequencies should be inaudible to anyone anyway, but cutting them away will prevent “bad” speakers from generating noise. This by the way also explains why it is named “speaker compensation”. Since (as stated in the article) RMAA doesn’t show any significant difference in the audible (sub-21.2kHz) region, I would recommend leaving this feature turned on, unless you know for sure you need the DPC-latency improvement.

    Reply

    • CarvedInside19 Oct 2017 @ 20:58

      Thank you for the information. This could be the its purpose. I am yet to read on low pass frequency filters and if other non-Xonar soundcards do something similar.
      If what you say is correct, those with very good hearing and good equipment would still benefit from turning “speaker compensation” off.

      Reply

      • dramu21 Oct 2017 @ 19:20

        > If what you say is correct, those with very good hearing and good equipment would still benefit from turning “speaker compensation” off.
        True. The problem is, only a very few people will satisfy those conditions, even if they think they do. Hearing above 20kHz is already exceptional unless you’re a teen, and hearing above 21.2kHz is even less likely. And in addition you need a good quality sound system. Recommending to turn this feature off will have negative consequences for 99 out of 100 people and will help only 1 (though a lot more will *think* it helps them).

        Reply

  6. Alex R19 Oct 2017 @ 00:28

    Hello. I’ve been intrigued by this tweak and followed the comment section since the start. I used to do a lot of audio editing for short films and mastering CD/Vinyl records as a hobby and for some additional cash. I performed on stage as an electronic artist and recorded few albums. From my years of experience after analyzing the spectrum I can say with 100% certainty that this tweak is a classic placebo. I know most of people here will get offended by reading this but placebo effect in audio industry is very common. As shown on FFT analysis this is purely a steep low pass frequency filter and does not affect other freqs (the small differences we see are just minor recording/playback artifacts). I can’t think of any other reason for implementing such filter in the drivers than for a bugfix or a workaround for a bug in the drivers (or like dramu said in his comment: to compensate for low quality speakers and prevent them from trying to reproduce inaudible frequencies what could in time degrade the speaker). That would explain why some people get performance problems after applying the tweak. I would recommend leaving that option on, I trust the developers on this one.

    Reply

    • Mescalamba19 Oct 2017 @ 18:38

      Its from DX card, which is very far soundwise from STX and such.

      Difference between no tweak and tweak on STX if ran thru WASAPI (and usual stuff in F2K) is pretty audible. Also requires probably set of good ears and headphones, thankfully I own both.

      Plus difference on graph might be rather small, but overall effect on sound might be bigger. Either way, output from tweaked version is decent bit cleaner than without it.

      Reply

      • Alex R19 Oct 2017 @ 23:12

        It’s a placebo until there’s any evidence to the contrary. Anyway tweak in a driver is a tweak in a driver. Sound card does not matter in this kind of test because you don’t want to test your sound card or your headphones or even your ears.

        Reply

        • Bill Gates20 Oct 2017 @ 20:11

          It’s not placebo. Quality difference is very big and statistically evident in blind abx tests. BTW, sound have many parameters, apart from spectrum.
          If you cant hear difference, maybe you have some ear damage, because of performing on stage, or it is age-related. But it is definitely something wrong with you.
          Also, I’m audio professional, audio and mastering engineer, editor, musician, director and post-production (multi-channel also) editor.

          Reply

        • Evgenim20 Oct 2017 @ 20:34

          No placebo at all. Difference is really hearable on D1.

          Reply

          • Alex R21 Oct 2017 @ 13:02

            Bill Gates sound has only three main parameters which can be easily measured. It’s overall loudness, differences in spectrum image and stereo image. And this tweak does not change the stereo image and FFT test shows it does not change the level either. It only applies a low pass filter so I have no idea what kind of audio voodoo you hear with your magic ears 😉 Without any evidence your opinion is still just a placebo effect. Audio is maths and your subjective perception of the sound doesn’t solve the meaning of this tweak.

            Reply

            • 3e7421 Oct 2017 @ 17:05

              I think we all agree that the placebo effect is very common in audio comparisons, and everyone likes to dismiss it in his/her own case. However, on my second edition STX1 with OPA1612 opamps, connected to Tannoy Reveal 402s using Klotz AC110 cabling, I can clearly hear the difference. Using headphones the difference is more subtle, but still hearable.

              Reply

            • Bill Gates21 Oct 2017 @ 23:45

              Alex R first of all, you must teach what is blind abx test. If abx test says that difference is confirmed, then it is not a placebo. Abx test is objective evidence. You conclusions are unscientific and completely ignoring abx test, what means that you are biased in the opinion and simply you muddle as it is convenient to you.
              “It’s overall loudness, differences in spectrum image and stereo image.” – is is nonsense. Sound has much more parameters. You can capture audio signal in digital form bit-to-bit from the dac of audio card and never see the same picture, if you sum signals in anti phases. Even if both signal are without this tweak.
              Sorry, but your argument look to very childish, superficial, completely unscientific. I think you’re not audio editor or musician, just a troll which ignore uncomfortable for you rational arguments ant talks same useless nonsense just for flame.

              Reply

              • Alex R04 Nov 2017 @ 11:59

                I know sound has many more minor parameters, but these are the main ones that define the structure of sound and only these are needed to write it into a file and reproduce it. You are probably talking about minor creative parameters that fall under those main categories. Why would you want to sum signals in anti phases? It’s not relevant to this driver tweak and you know it so stop calling me a troll since all you do is troll. Where is the documentation of abx tests? Anyway abx tests are not always accurate, you would need a perfect studio environment and a bit more people than you and your mate to take the test. So far the only proof we have is showing that this is a placebo or a high frequency boost for people with superhuman hearing.

                Reply

                • Bill Gates08 Nov 2017 @ 16:38

                  Alex R – “So far the only proof we have is showing that this is a placebo” – No proof of that from your side. Only in your failed brain. But the proof from all other people, that this is not placebo and real sound quality rise. Also ABX test with 100/100 result which is accurate evidence with 99.99^-99% of probability of a mistake.
                  “It’s not relevant to this driver tweak and you know it” – it is relevant to verification of sound difference and as consequence – to the quality. You know it but you continue to behave as the failure troll.
                  “but these are the main ones that define the structure of sound and only these are needed to write it into a file and reproduce it.” – nope. You don’t even know, that your primitive tests “stereo image and FFT test” uses so rough approximations (at least 2048 samples for one visual image, but of course you dont know what it is), that you cant use them as SQ test. The fact that you use such rough instruments and name them “maths” tells everything about you – and you continue to be either very not smart (and, as consequence, not understanding this), or a failed troll.

                  Reply

  7. dramu21 Oct 2017 @ 19:09

    Can somebody who hears a difference do a proper loudness measurement? I mean not with their ears, but with equipment. What I’m thinking of, it is possible that this tweak alters loudness in a subtle way. That would explain why there is no difference in frequency plots, yet people think there is. Because people subconsciously perceive louder audio as better quality, even if there is no actual difference (other than the loudness). This is true even if the difference in loudness is hardly noticable.

    Reply

  8. Johnyn06 Nov 2017 @ 20:12

    Any update on this? Is it an improvement? Where does the SpeakerCompensation setting come from? Windows 10 or Asus?

    Reply

    • CarvedInside07 Nov 2017 @ 01:16

      At the moment I don’t have anything else to say. You should try it and see if you consider it being an improvement. This is a Xonar driver setting.

      Reply

  9. D09 Nov 2017 @ 15:22

    Thx for the driver and your continuous support maybe the filter (the untweaked driver applies) has to do something with this
    https://people.xiph.org/~xiphmont/demo/neil-young.html
    under the “192kHz considered harmful” paragraph:
    “Neither audio transducers nor power amplifiers are free of distortion, and distortion tends to increase rapidly at the lowest and highest frequencies. If the same transducer reproduces ultrasonics along with audible content, any nonlinearity will shift some of the ultrasonic content down into the audible range as an uncontrolled spray of intermodulation distortion products covering the entire audible spectrum. Nonlinearity in a power amplifier will produce the same effect. The effect is very slight, but listening tests have confirmed that both effects can be audible.”
    It also has some pretty interesting tests right there.

    Reply

    • Mescalamba09 Nov 2017 @ 18:39

      Probably not, cause thats related to having 192kHz source. I doubt anyone has 192kHz source. Anyway, I have 44/48k and upsample usually to 2x. Difference is there constantly.

      Its just some built-in filter that cuts out details for whatever reason. Could be even bad programming. Its not like ASUS drivers are best at anything.

      Oversampling isnt problem (thats what resamplers do and why we have them).

      Reply

  10. Dima19 Nov 2017 @ 10:40

    Placebo or not, the sound is more clear to my ears. I know my crappy speakers inside and out because i create music with them for the last 3 years.

    Reply

AdminAuthorFeaturedBuried

Leave a Reply

Your comment may be held for moderation. If it does not show up immediately, just be patient. Comments have to comply with these rules.