• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

mansr

Major Contributor
Joined
Oct 5, 2018
Messages
4,685
Likes
10,717
Location
Hampshire
If you are talking about the potential for inter-sampling peaks, this is not an issue in current DACs. From what I understand, they typically keep sufficient headroom internally with floating point precision for any such peaks.
Whatever gave you that idea? Benchmark and RME seem to be the exceptions here.
 

Biblob

Addicted to Fun and Learning
Joined
Sep 13, 2018
Messages
637
Likes
607
I don't agree with the blanket recommendation of reducing source by 4dB (for an edge case that may or may not exist in your particular chain). The attenuation is not free. It has implications on dynamic range and if you are using a marginal amp later on with very efficient speakers, it might make the difference between hearing or not hearing hiss with normal volume. Combine that with the headroom you may lose for any room eq in the chain, the situation gets worse. So, the solution may be worse than the problem especially if the problem does not exist in your chain!

The correct solution is for the DAC (as a device not chip) vendor to take care of this depending on the chip they use and what the chip does. They can compensate in the gain stage later so that any attenuation they do pre can be normalized back to spec output voltage while keeping the noise components under control.

Do you know of any well-measuring currently available DAC that suffers from this?
Could I trust that my Asus Xonar U7 does this as well?
 

Vasr

Major Contributor
Joined
Jun 27, 2020
Messages
1,409
Likes
1,926
We can turn this around , do you know of anyone besides benchmark that actually cares about this why do you assume other manufacturers cares about it and have a fix implemented ?
That is a good question. Perhaps better asked of vendors like Topping, etc., as to what their view on it is. My point is whether manufacturers address it because it is a real problem or ignore it because it has very little practical significance doesn't matter to me. But doing a prophylactic -4dB attenuation that it might happen is like cutting off your hand to prevent a possible cut finger while using a knife.

Clearly Benchmark is "talking their book" as a differentiating feature. Are they hyping something that is a non-issue in practice? That is the real question.

It was even a problem with filters in CD players long ago and a known issue long before benchmark wrote some white paper about it .
First of all, this problem is being overstated.

To understand, this only applies to content that has "over 0dB" peaks but which is hidden because it falls between two samples. Lower the sampling rate, greater the chances and higher the potential peak overshoot that may be hidden.

This problem won't happen if it is soft-clipped (as happens with over-boosted recordings) with a sample point capturing a soft-clipped value at 0dB or less.

Also 3dB-4dB is the extreme case.
It's further compounded by our wish to also use EQ and room correction , so they should care that's rigth , but I'm not seen many that care.
No, using room EQ correctly will not create this problem just because of its use. Room EQ reserves head room for any boost it may provide. It might even digitally-clip if that head room isn't correct but the sampling if it captures that clipping will faithfully reproduce it in the DAC, not create hidden inter-sample peaks.
But this precautionary attenuation adds on to lower dynamic range further from the headroom reserved for room eq making it worse.
Is it in Amirs test suite to test for problems with intersample overs , or can it be inferred from what's already measured ?
I suspect it would not be because calibrated signals used as input would not have "over 0dB" hidden peaks between samples. The "implied" peaks would be at 0dB.

I still think people doing this blanket 4dB attenuation may be doing it unnecessarily and with a harmful impact on their audio.
 

bigguyca

Senior Member
Joined
Jul 6, 2019
Messages
483
Likes
622
- Omitted -

To understand, this only applies to content that has "over 0dB" peaks but which is hidden because it falls between two samples. Lower the sampling rate, greater the chances and higher the potential peak overshoot that may be hidden.

- Omitted -


What is the source of these intersample overs that exceed 0 dBFS?
 

restorer-john

Grand Contributor
Joined
Mar 1, 2018
Messages
12,955
Likes
39,716
Location
Gold Coast, Queensland, Australia
First of all, this problem is being overstated.

Totally.

Benchmark are experts in the fine art of overstating and discovering problems that either:
a) are insignificant
or
b) have been identified and solved decades ago. Look at how much they overstated the crossover distortion "issue" they also apparently had "solved". :facepalm:

I still think people doing this blanket 4dB attenuation may be doing it unnecessarily and with a harmful impact on their audio.

What's 4dB between friends? ;)
 

Vasr

Major Contributor
Joined
Jun 27, 2020
Messages
1,409
Likes
1,926
What is the source of these intersample overs that exceed 0 dBFS?

Bad, amateur or overzealous digital mastering. To keep it as loud as possible, they might push the input levels to as high as they can before they get clipping indicators. As Benchmark points out correctly, that may be subject to two successive sampling points being at 0dBFS or less (so no clipping indicator) while the peak is actually between them above 0dBFS and unsampled and so not represented in the digitized signal. But when a DAC tries to interpolate between the two sample points, it might reconstruct the missing content between two sample points which will be above 0dBFS.

So, it requires a confluence of a couple of things to happen and certainly not as prevalent as Benchmark seems to suggest.
 

HammerSandwich

Major Contributor
Forum Donor
Joined
Nov 22, 2018
Messages
1,137
Likes
1,506
I still think people doing this blanket 4dB attenuation may be doing it unnecessarily and with a harmful impact on their audio.
Fair point, but if your DAC/amp combo clips below full scale...

Let's consider some numbers. A power amp with 26dB gain increases voltage by 20x. If it clips at 2Vrms input, it outputs 40Vrms, or 200W into 8-ohm speakers. With 29dB gain or 100W max, you clip at 1.4Vrms from the source, and a -3dB limit costs nothing from most DACs, right? With 29dB gain and 100W, you must attenuate the digital signal or clip.

Rescale to your hardware.
 

Blake Klondike

Senior Member
Forum Donor
Joined
Jan 20, 2019
Messages
448
Likes
314
As someone who appreciates this thread but doesn't understand any of it, may I ask for a bottom line summary? Do I need to worry about audibly compromised sound on a windows PC unless I perform some arcane procedures?
 

Restin

Member
Joined
Dec 31, 2020
Messages
6
Likes
3
One quick question. Which is better sound quality: Windows at 100% volume or 25% volume with speaker gain increased. I read that the Windows setting to higher volume gives more bit depth, is that true?
 

xykreinov

Senior Member
Forum Donor
Joined
Nov 16, 2019
Messages
424
Likes
680
These issues are bypassed if you use waspish or ASIO
Ya, that's what I do when on Windows regardless of sound quality. I just prefer the lower latency. Though, high latency is inherent to multi-program audioprocessing from Windows' DirectSound, Linux's PulseAudio, etc.
 

Vasr

Major Contributor
Joined
Jun 27, 2020
Messages
1,409
Likes
1,926
As someone who appreciates this thread but doesn't understand any of it, may I ask for a bottom line summary? Do I need to worry about audibly compromised sound on a windows PC unless I perform some arcane procedures?
One quick question. Which is better sound quality: Windows at 100% volume or 25% volume with speaker gain increased. I read that the Windows setting to higher volume gives more bit depth, is that true?

Simple answer:
1. Use ASIO or WASAPI exclusive mode if your chain supports it for your application and use cases
2. If shared mode audio is being used
a. Select a sample rate for the audio device (in Windows Sound Devices) that is the highest of the most common content you play
b. Set the volume level for the audio device (in Windows Sound Devices) to about -0.2db (Hint: right click on the volume level allows you to switch between db scale and 0-100 scale).

Enjoy.

If you have OCD, follow the recommendations in this thread.
 

Vasr

Major Contributor
Joined
Jun 27, 2020
Messages
1,409
Likes
1,926
Fair point, but if your DAC/amp combo clips below full scale...

Let's consider some numbers. A power amp with 26dB gain increases voltage by 20x. If it clips at 2Vrms input, it outputs 40Vrms, or 200W into 8-ohm speakers. With 29dB gain or 100W max, you clip at 1.4Vrms from the source, and a -3dB limit costs nothing from most DACs, right? With 29dB gain and 100W, you must attenuate the digital signal or clip.

Rescale to your hardware.

If you have the luxury of explicitly gain-staging your chain, you can arrange for it to have the peaks correspond to the max output to the speakers with or without attenuation (albeit at the expense of S/N and/or dynamic range relative to the content). This is seldom a practical problem except when DAC/Pre-amp and Amp or badly matched or people play at max volume. Most reputable amps also have some (limited) head room above stated input sensitivity before they will clip.

The more common outcome of self-attenuation ;) is a lower volume at any listening level which if compensated for can increase the relative noise component. It is like attaching a 1.4Vrms output source instead of a 2.0Vrms output source for an amp that has 2.0Vrms input sensitivity. Nothing clips here.

Of more interest (in the edge cases) is what the DAC does if and when it finds inter-sample peaks. Does it clip (soft, hard)? Does it output more than its stated output voltage for such peaks (the amp may still not have a problem with this within limits)? Or does it behave badly and introduce artifacts? Real-life evidence for audible manifestations of this is not very strong.
 

Blake Klondike

Senior Member
Forum Donor
Joined
Jan 20, 2019
Messages
448
Likes
314
Simple answer:
1. Use ASIO or WASAPI exclusive mode if your chain supports it for your application and use cases
2. If shared mode audio is being used
a. Select a sample rate for the audio device (in Windows Sound Devices) that is the highest of the most common content you play
b. Set the volume level for the audio device (in Windows Sound Devices) to about -0.2db (Hint: right click on the volume level allows you to switch between db scale and 0-100 scale).

Enjoy.

If you have OCD, follow the recommendations in this thread.
I understand ASIO to be a sound option in audio interfaces-- is there a way to use it as a global system choice?
 

bennetng

Major Contributor
Joined
Nov 15, 2017
Messages
1,634
Likes
1,693
Keep in mind that the 0.2dB "recommendation" to avoid CAudioLimiter is relative to 0dBFS signal, the highest possible level for fixed-point signal. Audio players with floating point lossy decoders can have untouched decoded data beyond 0dBFS, as illustrated in this article:

http://archimago.blogspot.com/2019/06/guest-post-why-we-should-use-software.html

Search for the phrase "Realize that even if we don't use any DSP features" for the relevant part.

As for another "recommendation" of 4dB attenuation to prevent intersample overs, read the paragraph right above that, starting with "It is necessary to understand".

For more information, read the whole article from start to end.

With ReplayGain it is possible use different playback level adaptively for different kinds of music, something a fixed global volume control cannot achieve.
 

bennetng

Major Contributor
Joined
Nov 15, 2017
Messages
1,634
Likes
1,693
Benchmark are experts in the fine art of overstating and discovering problems that either:
a) are insignificant
or
b) have been identified and solved decades ago. Look at how much they overstated the crossover distortion "issue" they also apparently had "solved".
Well, yes. Because a 30+ years old Discman and 20 years old CD player had intersample headroom as well.

https://www.audiosciencereview.com/...in-intersample-overs-please.11651/post-338342
https://www.audiosciencereview.com/...in-intersample-overs-please.11651/post-401439
 

mugbot

Active Member
Forum Donor
Joined
Jul 3, 2020
Messages
120
Likes
110
Location
Adelaide
These solutions use ASIO or WASAPI, both of which prevent the user of equaliser APO. Is there a solution for quality windows audio with EQ APO enabled?
 

mugbot

Active Member
Forum Donor
Joined
Jul 3, 2020
Messages
120
Likes
110
Location
Adelaide
Top Bottom