Ending the Windows Audio Quality Debate

[Offler]

Did you match the channel count in the playback software, recording software, windows settings, and third party driver settings? Just want to tell you I don't have this issue, and therefore not universal to everyone.

Yeah, my point was that the problem with CAudioLimiter and audibility of the distortion IS probably linked to my specific setup.

Just did some testing of Windows sample rate converter. Recording (lineIn) is set to 192KHz, FLOAT.

I generated dual sine at -6,15dBFS so the issues with CaudioLimiter do not trigger for me. One frequency was 18000Hz, the other 19500Hz. This screen shows just noise spectrum, without any signal:

Here, the sines generated through in 44.1KHz sample rate and windows output is 44.1KHz 24bit:

1:1, no conversion was being used.

Signal has 44.1Khz sample rate, and windows output is 48Khz.

There you can see some peaks at 300Hz, 1200HZ, around 10 and 20Khz.

And then i tried 44.1KHz signal samplerate with 192KHz 24bit windows output:

There are still some peaks, but most of them are above 20Khz.

Then I tried 48KHz signal samplerate, here on 44.1Khz 24bit output:

So far worst case scenario.

48KHz signal to 48Khz 24bit output:

Almost identical to 44.1Khz on 44.1...

48KHz signal to 192Khz 24bit output:

And again you can see some peaks above audible spectrum.

This is just to demostrate that Windows 10 on my particular combination of ALC1220, NAD d3020v2 over toslink does handle sample rate conversion differently from practically inaudible to quite bad.

 

[bennetng]

Yeah, my point was that the problem with CAudioLimiter and audibility of the distortion IS probably linked to my specific setup.

Just did some testing of Windows sample rate converter. Recording (lineIn) is set to 192KHz, FLOAT.

I generated dual sine at -6,15dBFS so the issues with CaudioLimiter do not trigger for me. One frequency was 18000Hz, the other 19500Hz. This screen shows just noise spectrum, without any signal:

View attachment 162377

Here, the sines generated through in 44.1KHz sample rate and windows output is 44.1KHz 24bit:

View attachment 162378
1:1, no conversion was being used.

Signal has 44.1Khz sample rate, and windows output is 48Khz.
View attachment 162379
There you can see some peaks at 300Hz, 1200HZ, around 10 and 20Khz.

And then i tried 44.1KHz signal samplerate with 192KHz 24bit windows output:
View attachment 162380
There are still some peaks, but most of them are above 20Khz.

Then I tried 48KHz signal samplerate, here on 44.1Khz 24bit output:
View attachment 162381
So far worst case scenario.

48KHz signal to 48Khz 24bit output:
View attachment 162383
Almost identical to 44.1Khz on 44.1...

48KHz signal to 192Khz 24bit output:
View attachment 162384
And again you can see some peaks above audible spectrum.

This is just to demostrate that Windows 10 on my particular combination of ALC1220, NAD d3020v2 over toslink does handle sample rate conversion differently from practically inaudible to quite bad.

How about always use 96k for both analog and digital output, and only test input sample rates with test files <= 96k?

[EDIT] Do you have these settings available?

 

[Offler]

How about always use 96k for both analog and digital output, and only test input sample rates with test files <= 96k?

[EDIT] Do you have these settings available?
View attachment 162466

Yes, i have this settings available. I use only digital output

96KHz samplerate with 44.1KHz signal:

Apparently there is still some distortion related to conversion, which is not present on 192KHz...

Using 192KHz for shared mode would not be much of a problem in term of audio quality as i would have expected. Option to not use any sample rate conversion is still the best in terms of quality, but only by a slight margin in quality and performance, since shared mode already uses a lot of compressed audio - web applications, streams and games. Critical listening - CDs, BDs, DVDs, FLACs can be still ran in exclusive mode.

 

[bravomail]

Yes, i have this settings available. I use only digital output

96KHz samplerate with 44.1KHz signal:
View attachment 162758
Apparently there is still some distortion related to conversion, which is not present on 192KHz...

Using 192KHz for shared mode would not be much of a problem in term of audio quality as i would have expected. Option to not use any sample rate conversion is still the best in terms of quality, but only by a slight margin in quality and performance, since shared mode already uses a lot of compressed audio - web applications, streams and games. Critical listening - CDs, BDs, DVDs, FLACs can be still ran in exclusive mode.

just pick you poison - 44.1 or 48KHz sampling rate. I use WASAPI - so don't care. But for games, and browser Youtube - you can do 48KHz. That's how most of their material encoded.
BTW, there is an app SMPlayer, which will output Youtube audio via WASAPI!
VLC movie player will also do WASAPI for u.
That leaves games. They r all 48KHz. Problem solved!

 

[BillyD]

This seems to be great I love the sound but I have a couple questions.

1) I can't disable audio enhancements and have EAPO operating at the same time. If I install EAPO audio enhancements get enabled automatically and if I disable them it stops working. So what do I do here just keep them on? In the OP it says to disable them so I am confused.

2) In the APO configurator I can uncheck Use original APO for Pre-Mix but for Post-Mix it's greyed out. Why?

 

[filipduarte]

This seems to be great I love the sound but I have a couple questions.

1) I can't disable audio enhancements and have EAPO operating at the same time. If I install EAPO audio enhancements get enabled automatically and if I disable them it stops working. So what do I do here just keep them on? In the OP it says to disable them so I am confused.

2) In the APO configurator I can uncheck Use original APO for Pre-Mix but for Post-Mix it's greyed out. Why?

1) You need to disable the enhancements on the "Enhancements" tab if your device has it.
On the "Advanced" tab, under the "Signal Enhancements" section, you must check the "Enable audio enhancements" box to EAPO work.

2) You can ignore the "Use original APO" boxes that are grayed out.
The "grayed out" signifies that your device's original driver didn't have custom APOs.
But if you can uncheck, do it.

 

[Uninen]

1) You need to disable the enhancements on the "Enhancements" tab if your device has it.
On the "Advanced" tab, under the "Signal Enhancements" section, you must check the "Enable audio enhancements" box to EAPO work.

View attachment 167910View attachment 167911

2) You can ignore the "Use original APO" boxes that are grayed out.
The "grayed out" signifies that your device's original driver didn't have custom APOs.
But if you can uncheck, do it.

Does anyone know if something has changed with Windows or EAPO because for me, the EAPO filters work whether the "Enable audio enchantments" is on or off. I haven't critically listened to it but when I disable it the sound seems to get a tiny bit louder. And switching filters / preamp gain still affects the sound and volume. When enabling it the sound level goes down a tiny bit and the EAPO still affects the sound. There is a delay and silence of maybe 1 second when switching.

So could it it be a bug? Or does EAPO nowadays "go deeper" to Windows drivers and it isn't affected by that selection?

The I downloaded the EAPO maybe 2 days ago so it should be the latest and I'm using Win10.

 

[Offler]

just pick you poison - 44.1 or 48KHz sampling rate. I use WASAPI - so don't care. But for games, and browser Youtube - you can do 48KHz. That's how most of their material encoded.
BTW, there is an app SMPlayer, which will output Youtube audio via WASAPI!
VLC movie player will also do WASAPI for u.
That leaves games. They r all 48KHz. Problem solved!

Unfortunately no. It seems that games and Youtube are nowadays still using a mixed bag of 44.1 and 48KHz content - I even have a player which can play youtube videos in exclusive mode and show their stats. Unless you spend some time examining, you would never know in what samplerate is the sound encoded.

I dont see much benefits of high-res audio (like anything above 44.1/48KHz 16bit maybe 24bit) in terms of records, but using 192KHz 24bit as an windows default shared output just for negating aliasing effects from all the resampling and mixing makes sense.

Long story short - if you want to get good speakers with good amplifier for Windows system, you definitely can. Just keep in mind that getting really good results might take some time and effort.

 

[edechamps]

Hello there,

I'm trying to answer a question that is somewhat related to this thread.

A bit of background: most audio device/drivers advertise a hardware volume control, such that the main Windows volume control simply tells the hardware device what to set the volume to. However, some audio devices do not expose any hardware volume control, and in that case, Windows automatically inserts a digital volume control into the software processing chain.

What I'm curious about is: if that software volume control is used, and the volume is set to less than 100%, is it possible to have an APO (e.g. Equalizer APO) output more than 100% signal level without clipping? For example, say the volume control is set to -6 dB, and I set a gain of +3 dB in some APO, and the application sends a full scale sine wave, will that sine wave clip or be limited by CAudioLimiter? Or is the hardware device simply going to receive a -3 dBFS sine wave without clipping?

The reason I'm wondering about that is because, if it doesn't clip, then such a behaviour could be quite interesting for people using, say, Equalizer APO for room EQ, because it would simplify gain staging and make it easier to use the full dynamic range of the system. In my case it would allow me to get rid of the whooping -12 dB anti-clipping attenuation I currently have set on my room EQ.

Unfortunately it looks like all the DACs I have lying around have hardware volume control so I can't use them to test my theory. Intuitively, I would think that the typical case where the device is missing hardware volume control is where there is no D-A conversion going on - i.e. digital-to-digital converters (e.g. S/PDIF outputs). But I don't really have those at hand.

So, I have a request to those using devices with no hardware volume control if they would be so kind to check this for me.

How do I know if Windows is using hardware volume control with my device? Simple: does the master Windows volume control work in WASAPI Exclusive Mode? If yes, you have hardware volume control. If not, Windows is using software/digital volume control and you can contribute to this little experiment!

Okay, I have software volume control. What should I do? Install Equalizer APO on the device and set up a +10 dB preamp/gain. Set the Windows device volume to -20 dB (or less). Then output some full scale (close to 0 dBFS) sine wave through the device using the software of your choice (e.g. REW). Finally, switch the preamp/gain filter on/off repeatedly in Equalizer APO while the sine wave is playing. Listen to the sine wave. Does it get much louder when you turn the filter on? (Try it with a -10 dBFS sine wave first if you're not sure Equalizer APO is working.)

I'd be very interested in your results! Please also mention the model of audio device you are using and whether it's a DAC or a digital-to-digital converter.

 

[Offler]

First you can look at the construction of the soundcards.

Mine is Realtek ALC1220, which is just "plain" DAC. Most of the work is done by CPU that includes all the processing. So everything from the start to the end is done by software. Any sort of volume control is basically just attenuation, but no gain.

When WASAPI: Exclusive over TOSLINK is engaged there are exactly two ways how the sound goes out:
a) On its original loudness.
b) Completely muted.

The only volume control I can use is on the external amplifier or player software (Foobar, MPC), but Windows Volume does not do anything.

Then there are products like this one:

Hi-Fi｜Sound Cards｜ASUS Global

Xonar AE is a PCIe gaming sound card built with high-quality components to satisfy gamers who demand exceptional sound for the best gaming and multimedia experiences. Xonar AE delivers 7.1-channel, 192kHz/24-bit Hi-Res audio output with a high 110dB signal-to-noise ratio (SNR), and comes with a...

www.asus.com

Because its for both PC crowd and for audiophiles, they provided a lot of technical details. TI Bur-Brown PCM1792A, Asus audio processing engine and TI TPA6210A2 amplifier...

The amplifier is placed after DA conversion (at least it seems so), and i would assume it might contain some sort of volume control. The controls might be available only in driver/proprietary SW, or as you posted through OS, or it might be set to amplify the sound by a fixed value, without any control from the outside.

So if you have hardware amplifier, i would assume it acts exactly as an external one with a volume knob. There can be real volume gain.

There is also a SPDIF output over coaxial connector. Volume control for it can work through Audio Processing Engine, and again this can be made visible in OS - either through volume control, or proprietary audio control panel.

But technically speaking... the driver and the hardware on the card can do all the processing. Decompression, mixing... There are different limitation than to a Windows Audio Stack, but there are some. APO are in such case completely useless, its like when you send audio over Wasapi exclusive to a device which performs all other tasks.

To answer your question regarding software volume control:

- Master Volume on 100%
- Gain in EAPO is switching between 0.00 and +10.00
When the +10db filter is enabled, volume does not increase. When the filter is disabled, volume goes more silent for a while, then returns back to original value.

- Master Volume on 50%
- Gain in EAPO is switching between 0.00 and +10.00
When the +10 filter is enabled, volume does increase, both audibly and measurably. When its disabled, volume goes returns to lower value.

There is apparently a relation between EAPO and Master Volume control in Windows. When master volume is on 100%, EAPO has to be set at -0.14dB so CAudiolimiter will not trigger/distort sound for single source, or -6.15 for two.

 

[edechamps]

There is also a SPDIF output over coaxial connector. Volume control for it can work through Audio Processing Engine, and again this can be made visible in OS - either through volume control, or proprietary audio control panel.

But technically speaking... the driver and the hardware on the card can do all the processing. Decompression, mixing... There are different limitation than to a Windows Audio Stack, but there are some. APO are in such case completely useless, its like when you send audio over Wasapi exclusive to a device which performs all other tasks.

To answer your question regarding software volume control:

- Master Volume on 100%
- Gain in EAPO is switching between 0.00 and +10.00
When the +10db filter is enabled, volume does not increase. When the filter is disabled, volume goes more silent for a while, then returns back to original value.

- Master Volume on 50%
- Gain in EAPO is switching between 0.00 and +10.00
When the +10 filter is enabled, volume does increase, both audibly and measurably. When its disabled, volume goes returns to lower value.

There is apparently a relation between EAPO and Master Volume control in Windows. When master volume is on 100%, EAPO has to be set at -0.14dB so CAudiolimiter will not trigger/distort sound for single source, or -6.15 for two.

Interesting! That would indeed seem to confirm my suspicion that, if Windows uses its own software volume control, then an APO can output samples "beyond" the normal range that would normally be possible with hardware volume control (if the control is set to less than 100% of course). Which does make sense given that the pipeline is floating point, and if we assume that CAudioLimiter acts after the software volume control.

So if I understand correctly, with your Realtek ALC1220, Windows uses hardware volume control on the analog outputs, and software volume control on the S/PDIF output? That would be consistent with what I'd expect.

 

[Offler]

So if I understand correctly, with your Realtek ALC1220, Windows uses hardware volume control on the analog outputs, and software volume control on the S/PDIF output? That would be consistent with what I'd expect.

Unfortunately, the answer would be 'no'.

Even when there is headphone amplifier onboard on AsRock x399M next to ALC1220 DAC, analog output works in same manner as SPDIF in terms of volume control in WASAPi: Exclusive, but it seems that analog audio is louder by about 4dB. I would assume its just set for a fixed value and completely invisible to OS.

Also after reading some stuff about Microsoft Audio, Volume Control in Windows is also just an APO, just the one that cannot be usually bypassed in Shared mode.

But knowing that there are PCI-E cards which can bypass most of Windows Audio stack, and mainly APOs is quite interesting.

 

[edechamps]

Unfortunately, the answer would be 'no'.

Wait, I'm confused now… so you're saying Windows uses software volume control for all ALC1220 outputs (analog and digital)? Or none of them?

I also have a Realtek at hand - a ALC1220P-VB2 on a MSI MPG Z390 GAMING PRO CARBON. I don't have the tools to check its S/PDIF output, but I did check its analog outputs, and can confirm they use hardware volume control, as evidenced by the Windows volume control working in WASAPI Exclusive. That's true for my Xonar U7 as well, and more generally I would expect the same result on most USB DACs.

Also after reading some stuff about Microsoft Audio, Volume Control in Windows is also just an APO, just the one that cannot be usually bypassed in Shared mode.

That's my understanding as well when the driver does not advertise any hardware volume control to the OS. If it does advertise one, the APO is not inserted and the Windows volume setting value is sent to the hardware/driver instead, as documented by Microsoft themselves.

 

[edechamps]

I ended up checking all the DACs I could get my hands on. It appears they all use hardware volume control as the volume can be changed in WASAPI Exclusive mode for all of them Here's the full list of all the devices I tested:

• Realtek ALC1220P-VB2 on MSI MPG Z390 GAMING PRO CARBON (analog out)
• Apple USB-C dongle
• Some dirt-cheap noname USB DAC
• FiiO E7
• Xonar U3
• Xonar U7 (analog out)
• JDS Labs O2
• VIMVIP USB-C dongle (CHBT audio)
• Moshi USB-C dongle
• Google Pixel buds
• Corsair Virtuoso (both wireless and wired)

Still curious to hear if anyone with some kind of USB-to-digital (e.g. S/PDIF) interface can reproduce this. I'm especially interested in MiniDSP USBStreamer/U-DIO8 as I might end up buying one of those.

 

[Offler]

Wait, I'm confused now… so you're saying Windows uses software volume control for all ALC1220 outputs (analog and digital)? Or none of them?

I also have a Realtek at hand - a ALC1220P-VB2 on a MSI MPG Z390 GAMING PRO CARBON. I don't have the tools to check its S/PDIF output, but I did check its analog outputs, and can confirm they use hardware volume control, as evidenced by the Windows volume control working in WASAPI Exclusive. That's true for my Xonar U7 as well, and more generally I would expect the same result on most USB DACs.



That's my understanding as well when the driver does not advertise any hardware volume control to the OS. If it does advertise one, the APO is not inserted and the Windows volume setting value is sent to the hardware/driver instead, as documented by Microsoft themselves.

Unfortunately, many ALC1220 have vastly different implementation and its problematic to compare two 1220s. I was trying different drivers for mine, but the only one which worked properly was on AsRock webpage.

My understanding of Microsoft Web is the same, and I really would like to see it work.

 

[maverickronin]

It appears they all use hardware volume control as the volume can be changed in WASAPI Exclusive mode for all of them Here's the full list of all the devices I tested:

Still curious to hear if anyone with some kind of USB-to-digital (e.g. S/PDIF) interface can reproduce this.

So you're looking for a S/PDIF bridge where the volume control doesn't work when using WASAPI?

I have a couple of these, which I normally use to get digital audio out of my Nintendo Switch and into my RME ADI-2 DAC.

I just plugged one into my PC and the windows volume control didn't work when using WASAPI. It only does 16 bit at 32/44.1/48KHz though, so that might be a problem in your application.

 

[edechamps]

@maverickronin Interesting, thanks. The biggest constraint in my particular application is that I need at least 5 channels, hence why I was looking at MiniDSP's multichannel offerings.

I just plugged one into my PC and the windows volume control didn't work when using WASAPI.

If it's not too much hassle, it would be great if you could try the experiment I described in my original post:

Okay, I have software volume control. What should I do? Install Equalizer APO on the device and set up a +10 dB preamp/gain. Set the Windows device volume to -20 dB (or less). Then output some full scale (close to 0 dBFS) sine wave through the device using the software of your choice (e.g. REW). Finally, switch the preamp/gain filter on/off repeatedly in Equalizer APO while the sine wave is playing. Listen to the sine wave. Does it get much louder when you turn the filter on? (Try it with a -10 dBFS sine wave first if you're not sure Equalizer APO is working.)

 

[maverickronin]

If it's not too much hassle, it would be great if you could try the experiment I described in my original post:


edechamps said:

Okay, I have software volume control. What should I do? Install Equalizer APO on the device and set up a +10 dB preamp/gain. Set the Windows device volume to -20 dB (or less). Then output some full scale (close to 0 dBFS) sine wave through the device using the software of your choice (e.g. REW). Finally, switch the preamp/gain filter on/off repeatedly in Equalizer APO while the sine wave is playing. Listen to the sine wave. Does it get much louder when you turn the filter on? (Try it with a -10 dBFS sine wave first if you're not sure Equalizer APO is working.)

Have some brain fog today so I'm not exactly sure what you're trying to test. You want the test signal to be played through DS/WASAPI shared and see if EQ APO can raise the level above the Windows GUI volume control?

 

[edechamps]

You want the test signal to be played through DS/WASAPI shared and see if EQ APO can raise the level above the Windows GUI volume control?

That's basically it, yes.

 

[maverickronin]

That's basically it, yes.

I attempted to test this in a VM but I didn't see the DAC/bridge as a selectable device to pass though and then VirtualBox BSODed my physical machine.