Ending the Windows Audio Quality Debate

[Katji]

Bit more details on this page:
https://sourceforge.net/p/equalizerapo/wiki/Documentation/

Same page - I just used a specific page # place.

Very useful info but I don't know what to make of it.
Should I "uncheck only one of the Install Apo checkboxes" and see what happens?

See the configuration reference page too. https://sourceforge.net/p/equalizerapo/wiki/Configuration reference/
There are 2 pages - the author makes it confusing, makes it seem like there are more.

 

[Offler]

I don't understand. I see the small graph with the wave becoming smaller in amplitude, then high again. But what does it mean?

Is the big graph the sum of all the values recorded during the gain changing process, or an instant picture?

The FFT graph is sum of 10 measurements, but the upper one is from "mini" oscilloscope.

What happens is, that if I increase volume on 50Hz or 7000Hz above beyond -6dB, peak in the FFT graph will not raise. If i lower it to -6,15dB, the harmonic distortion will disappear. So this again indicate volume limiter, but it appears that in the center of the audible spectrum is close to 0, and at the edges is beyond -6db.

And before someone asks, yes, the distortion is audible from speakers.

Edit:
Just re-tested all frequencies with a single sine and then with two waves. For single wave it was -0,15dB, for two -6,15dB. Higher gain simply caused distortion. Unfortunately I dont have tools to test up to 32 tones.

 

[daftcombo]

The FFT graph is sum of 10 measurements, but the upper one is from "mini" oscilloscope.

What happens is, that if I increase volume on 50Hz or 7000Hz above beyond -6dB, peak in the FFT graph will not raise. If i lower it to -6,15dB, the harmonic distortion will disappear. So this again indicate volume limiter, but it appears that in the center of the audible spectrum is close to 0, and at the edges is beyond -6db.

And before someone asks, yes, the distortion is audible from speakers.

Edit:
Just re-tested all frequencies with a single sine and then with two waves. For single wave it was -0,15dB, for two -6,15dB. Higher gain simply caused distortion. Unfortunately I dont have tools to test up to 32 tones.

Can't you launch two instances of your two-tones generator, to have four tones playing? And so on...

Otherwise, with REW you can generate multiple tones. 32 should be possible.

Or just launch some music?

But what I don't understand is why the limiter would limit two tones to -6dB but one to -0.15dB.

 

[bennetng]

Been playing around with dual-sine and EQ APO and found this:

50Hz sinewave is having alot of loud harmonics which affected second sine at 7kHz.
View attachment 149449

So I opened EQ APO and started to lower the gain and at -6.30dB for both waves result was this:
View attachment 149450

I tried to repeat it several times... This happenned when I enabled EQ APO preamp at -6.3db. Wave at the small top graph first shinks at the exact moment when I lowered gain, and then gradually expanded. To me it looks like CAudio limiter. I am still testing it. (And no, I dont have any sort of Automatic Gain Control enabled)
View attachment 149452

I will probably try different gains for 50Hz sinewave.

You need to understand how to read FFT analysis, read the link below, very carefully:
https://www.audiosciencereview.com/...d90-balanced-usb-dac-review.10519/post-291483

That's why I said "don't use a sine wave" in this reply to @daftcombo
https://www.audiosciencereview.com/...indows-audio-quality-debate.19438/post-808237

I believe more to my eyes, than to my ears

Controlled, double-blind listening tests are completely fine, flawed measurement methodology on the other hand is bad.

Anyway, the title of the thread is more like a clickbait, better rename it to "EAPO tutorial/discussion" or something similar.

 

[Offler]

You need to understand how to read FFT analysis, read the link below, very carefully:
https://www.audiosciencereview.com/...d90-balanced-usb-dac-review.10519/post-291483

That's why I said "don't use a sine wave" in this reply to @daftcombo
https://www.audiosciencereview.com/...indows-audio-quality-debate.19438/post-808237


Controlled, double-blind listening tests are completely fine, flawed measurement methodology on the other hand is bad.

Anyway, the title of the thread is more like a clickbait, better rename it to "EAPO tutorial/discussion" or something similar.

I do understand that real-life scenarios mostly dont go for maximum volume, and use more than one or two tones.

Just to be sure, I was able to hear the difference between distorted and normal dual sine scenario. At this point i dont know if to attribute it to software I use for tone generation, or Windows audio stack.

 

[daftcombo]

That's why I said "don't use a sine wave" in this reply to @daftcombo
https://www.audiosciencereview.com/...indows-audio-quality-debate.19438/post-808237

When I want to create a WAV file with REW containing a two tones sinewave, I am not allowed to create it if I don't put -1.67dBFS, which is calculated to be equal to 0dBFS (peak). Is that what you refer to?

 

[danadam]

But what I don't understand is why the limiter would limit two tones to -6dB but one to -0.15dB.

I didn't really follow the discussion, but -6 dB is almost half of the linear scale. If you have two sources, which are mixed, and you want to make sure that the result doesn't clip, then the most certain way is to limit those sources to less than -6.02 dBFS. With 4 sources it is -12.04 dBFS, with 8 sources, -18.06 dBFS, etc.

 

[daftcombo]

I played in Foobar the dual-tone generated as stated above.

With -4dB in EqAPO as a preamp, I get this:

But without preamp in APO, I get this:

And with a preamp of -0.2dB:

So -0.2dB is not enough for me, but -4dB is sufficient.

 

[daftcombo]

Here's a multitone without pre-amp, 0dBFS peak With a preamp of -1,7dBFS it looks the same.

With a -10dB preamp, it seems just a bit better:

 

[Offler]

I didn't really follow the discussion, but -6 dB is almost half of the linear scale. If you have two sources, which are mixed, and you want to make sure that the result doesn't clip, then the most certain way is to limit those sources to less than -6.02 dBFS. With 4 sources it is -12.04 dBFS, with 8 sources, -18.06 dBFS, etc.

Basically explains why would CAudioLimiter even trigger, when earlier was proven that it triggers at -0,12dBFS with single source. -6,30dBFS would be logical answer for two separate signals.

Unfortunately, if the limiter is triggered, clipping is avoided, however there is introduced a some distortion into resulting signal, reaching -60dB and in certain scenarios its audible.

 

[daftcombo]

Basically explains why would CAudioLimiter even trigger, when earlier was proven that it triggers at -0,12dBFS with single source. -6,30dBFS would be logical answer for two separate signals.

Unfortunately, if the limiter is triggered, clipping is avoided, however there is introduced a some distortion into resulting signal, reaching -60dB and in certain scenarios its audible.

That means that a multitone of 4 tones would trigger the limiter if each is above -12dBFS, right? and -30dBFS for 32 tones?
And with actual music, how many tones can reach peak level at the same times?

The -0.12dBFS advised by OP seems useless now...

 

[kidoman]

If it is true, sound quality improvement is to be expected with a -6.3dB preamp then!

Perhaps @DDF or @edechamps would have an idea already?

I quoted the wrong post when replying!!!

 

[bennetng]

I do understand that real-life scenarios mostly dont go for maximum volume, and use more than one or two tones.

That's not what I wanted to say. Did you really carefully read what I posted?
https://www.audiosciencereview.com/...d90-balanced-usb-dac-review.10519/post-291483

Just to be sure, I was able to hear the difference between distorted and normal dual sine scenario.

Again, not what I wanted to say. Not that I don't believe you can hear a difference, my point is your measurement methodology is flawed.

That means that a multitone of 4 tones would trigger the limiter if each is above -12dBFS, right?

It depends on the tones' frequency and phase. Download the attached files and have a look. In frequency domain (FFT) both files look the same but in time domain (waveform view) one file has a higher waveform peak.

 

[danadam]

And with actual music, how many tones can reach peak level at the same times?

Err... none? Unless you happen to have a single frequency at some point, which wouldn't be very musical, methinks
Here is spectrogram of 5 seconds of full scale 1 kHz tone and then 1 minute of "Daft Punk - Get Lucky":

If we "zoom-in" to the top 20 dB we see that besides our tone, nothing reaches 0 dBFS:

The -0.12dBFS advised by OP seems useless now...

Why? When you play music, you usually play it from a single source, don't you?

 

[daftcombo]

It depends on the tones' frequency and phase. Download the attached files and have a look. In frequency domain (FFT) both files look the same but in time domain (waveform view) one file has a higher waveform peak.

You are right. This is how iZotope analyses them:

Visually, spectrums look the same, waveforms have different amplitudes as you say.

 

[daftcombo]

Err... none? Unless you happen to have a single frequency at some point, which wouldn't be very musical, methinks
Here is spectrogram of 5 seconds of full scale 1 kHz tone and then 1 minute of "Daft Punk - Get Lucky":
View attachment 149497
If we "zoom-in" to the top 20 dB we see that besides our tone, nothing reaches 0 dBFS:
View attachment 149498

Why? When you play music, you usually play it from a single source, don't you?

Do you mean that I shouldn't care about the content of the music, and that it's the job of the sound engineer to put out a track that will never reach 0dBFS?

But OP also advises to put something like -4dB instead of -0.12dB if we plan to use filters (which will modify phase). This explains that.

 

[Offler]

Err... none? Unless you happen to have a single frequency at some point, which wouldn't be very musical, methinks
Here is spectrogram of 5 seconds of full scale 1 kHz tone and then 1 minute of "Daft Punk - Get Lucky":
View attachment 149497
If we "zoom-in" to the top 20 dB we see that besides our tone, nothing reaches 0 dBFS:
View attachment 149498

Why? When you play music, you usually play it from a single source, don't you?

Some audio tracks howevever can reach 0dBFS, which is a topic of "Loudness wars". Some audio engineers actually take the loudest part of the song and allign it with 0 during mastering. Not a good practice in my opinion...

On other hand, Windows does mix multiple sources and this method is what makes games work with windows and other sounds. If you use Windows PC with Default Audio Renderer -0,12dB can help to remove most of the distortion when playing music even from single source.

 

[Offler]

Again, not what I wanted to say. Not that I don't believe you can hear a difference, my point is your measurement methodology is flawed.

Methodology itself is OK, I just needed to understand that mixing needs to lower the amplitudes by roughly 6dB. Thats why I did not understood why was the distortion introduced yet again.

 

[danadam]

There may be some misunderstanding somewhere. The question was "with actual music, how many tones can reach peak level at the same times?". I said at most one but it wouldn't be musical. This answer doesn't mean that no song waveform ever reaches 0 dBFS. They often do, but when this happens, it is many tones at non-peak level, which sum up to peak level.

 

[bennetng]

You are right. This is how iZotope analyses them:

View attachment 149503
Visually, spectrums look the same, waveforms have different amplitudes as you say.

~6dB for mixing two mono waveform is the worst case scenario, which assumes the two waveform being mixed are completely in phase. On the other hand, if two waveform are completely out of phase, they will be cancelled (null).

Google "pan law" for more information. 0dB to -6dB pan law values are commonly used in DAWs, depend on different mixing needs.

Methodology itself is OK, I just needed to understand that mixing needs to lower the amplitudes by roughly 6dB. Thats why I did not understood why was the distortion introduced yet again.

Just see daftcombo's iZotope analyses of the example files I posted. They look like the same in FFT view, but the waveform peak values are vastly different, and therefore the methodology of look at individual FFT bins for peak level alignment is flawed:
https://www.audiosciencereview.com/...indows-audio-quality-debate.19438/post-890577