Let's develop an ASR inter-sample test procedure for DACs!

[Salt]

./By and use the vinyl of it and ISO is out of discussion\.

 

[tccalvin]

Yes. The reconstructed waveform can peak above the samples, and my figures showed that.

I'd suggest a simple numerical experiment. To make the math a little simpler I'll use a sampling rate of 1 (i.e. t = 0, 1, 2, 3 ,4 5, ...).

• You can use a spreadsheet to calculate: w = A sin( 2 π f (t + δt) ) for a number of t's.
• The usual illustrations use A = sqrt(2), f = 0.25 and δt = 0.5.
• Try a smaller A and a slightly different f, and see how many periods of the waveform it takes before you hit w > +/-1. There you have clipped samples.
• Think what the probability of not having clipped samples will be with real life signals.

Remember that most of the frequency contents in whatever we listen to are mostly of frequencies << fs/4 (for redbook CD fs/4 = 11 025 Hz).

Sorry. I don’t follow you.

What’s the point of doing this mathematical modeling to try and guess the probability of ISOs not being accompanied by clipped samples when we have empirical data right here?

If your point is that having ISOs means that the signal is also likely clipping, then someone can take the songs that have already been tested for inter-sample peaks (this time without oversampling) and quickly put them in Audacity. There’s a simple option in that software to find clipped samples.

If it’s indeed the case that most ISPs are accompanied by many clipped samples, we should expect to find a clipping fest in those songs. I doubt that will be the case, because it would mean that the mastering engineers didn’t know how to do their job, but I would be happy to be proven wrong.

EDIT: Typos.

 

[VientoB]

As someone said - it happens at 11 kHz and the first harmonic is a 22kHz... so:

a) is this really a SQ problem? Probably not at all...
b) no, its not technically "nice"! Beauty contest prizes for grabs...

Q: Could anyone detect a 3dB compression at 11kHz - what music contains such levels at such frequencies?

//

It's not compression it's distortion.

 

[JIW]

Hi everyone,

In my previous example, it was a 5512.50Hz at 67.5° phase shift. It creates an harmonic at 16537.5Hz if clipped. That’s why I like this test tone, also because it creates a peak level up to +0.69 dBFS “only” and in audio band. So I kind of expect all DACs to handle well at least this one.

And back to my example with 5512.50Hz on the Teac VRDS-25X, it obviously massively overloads its ASRC and generates non-harmonic distorsion components at -60dB from 1kHz to 3kHz. We can only hope for the rest of the music to hide that.

I agree the 11025kHz is an extreme case. But if not real, it could demonstrate resistance of a DAC when fed with hot records that have suffered from loudness war. In the end, the only intention here is to show the difficulty of a DAC to deal with 0dBFS+ levels.

I know my case is different because I like to use CD players, and after 40+ years of having bought so many CDs - a lot being too hot - I’m eager to know which player is likely to best handle them.

From my reviews of CD players here, I discovered a lot of members still use them as transports feeding an external DAC, which might not offer digital volume control from S/PDIF inputs prior to feed their ASRC/SRC/OS filters. I think it’s interesting to know.

In terms of audibility, I’ll play with the Teac and see how I can share some test files with the community. That requires time though and is a different beast to cope with. In the meantime, I’ll continue to test IS over resistance

Where in the signal path the DAC chip allows for volume control should be in the data sheet. Whether the DAC manufacturer uses that feature or implements their own is of course up to them.

The ESS ES9039Q2M and ESS ES9039PRO have the volume control before any oversampling for PCM. The AKM AK4499 does as well denoted by DATT.

Source: https://www.esstech.com/products-overview/digital-to-analog-converters/sabre-audiophile-dacs/

Source (Download link): https://media.digikey.com/pdf/Data Sheets/AKM Semiconductor Inc. PDFs/AK4499_Feb2019.pdf

 

[Salt]

The link without G link

 

[JIW]

The link without G link

Thanks. Couldn't figure it out. Edited my post.

 

[NTTY]

As someone said - it happens at 11 kHz and the first harmonic is a 22kHz... so:

It’s just a test. It can be done at 5512.50Hz too but generates an over of 0.69dB max, which is more interesting on my perspective.
It does not generates harmonic distorsion only.

 

[tccalvin]

I just did a test myself. Not anything conclusive but it confirms what we’ve been talking about.

This is Careful by Paramore. Kind of a worst case scenario. Very compressed hard rock.

Not really happy to see all this red, gotta admit. But if we run the “find clipping” analysis tool (default settings) we can see that it’s mostly single clipped samples that snuck past the limiter. The way the tool works by default is that it starts flagging clipped samples when it finds at least 3 consecutive ones and it stops when it finds 3 consecutive non-clipped samples.

Now we use 8x oversampling:

Yikes. Now we have introduced a lot more clipping due to ISOs.

EDIT: removed the last picture to avoid confusion. It wasn’t really relevant anyway.

Since I’m not an expert and I’m not familiar with testing I encourage everyone to give feedback. If I messed up somewhere I’ll edit the post.

 

[Scytales]

It’s just a test. It can be done at 5512.50Hz too but generates an over of 0.69dB max, which is more interesting on my perspective.
It does not generates harmonic distorsion only.

John Siau made a live demo about the added distortion other than harmonics in case of ISO in this video from about 45'41'' :

 

[NTTY]

@Scytales: Yep, I saw the overload of the OS filter and the high level ultrasonic output it generates. That is why I wanted to show another and different effect (which I believe to be overload of an ASRC) and distortion/noise generated in audio band, this time. So my example is very much different from the one of @John_Siau but from the same IS over test.

Cheers

 

[KSTR]

That's back to front thinking and really doesn't help the issue.

If every DAC maker decides overnight to handle 3dB overs "gracefully", what do you think will happen? Recording will increase in level yet again and we'll be needing another adjustment, this time requiring sophisticated signal waveform prediction and reconstruction in real time to account for bad recordings.

It's too hot recordings and unnecessary oversampling after the fact causing the "problem".

We had that argument many pages ago, don't you remember?
And you still don't seem to understand what inter-sample overs are. When you increase record level even more beyond today's worst-case loudness war examples, even to the point that you have basically 1-bit data (the final loudness war brickwall), the IS-overs don't increase in value by much and more importantly they never exist for extended periods, they can only occur between samples (hence their name). Sustained trains of IS-overs that would actually affect measured and perceived loudness cannot exist except for those very academic special cases discussed above (@fs/4 etc with proper phase).

Full scale levels from brickwall limiting have nothing directly to do with IS-overs, you can have full scale levels all the way without any IS-overs, and you can have ton's of IS-overs in otherwise well-recorded material.

To cut the long story short,it's a horror story,specially the metal-ish stuff,something like a 10% of these specially some (unknown to me) smaller ensembles.
And to listen to them is just that,the glare that pure,old clipping brings,it does not shout "IS-Overs",it shouts bad recording intended to play low and stick out as loud.
It makes 50 yo stuff of Deep Purple for example to look SOTA,despite the age.

This! The moment tons of IS-overs that when clipped (or worse) would be a real problem, the music is already broken otherwise so it doesn't matter much. Typically at least.

The real problem is the more or less occasional IS-over that is handled badly, like in a SRC chip, with music style and recording quality where small artifacts actually have a chance to be audible.

 

[tccalvin]

Did another test. This time with a properly done master.

Now THIS is conclusive evidence:

0 clipped samples. Tons of ISOs. Clipping here is entirely created by oversampling.

In this case, the probability of having ISOs without any clipped samples is 100%, wouldn’t you agree?

 

[KSTR]

0 clipped samples. Tons of ISOs. Clipping here is entirely created by oversampling.

Good example. But we must be aware that this is valid only for software resampling in integer domain (Audacity is using the SoX library, AFAIK) where the values between the samples cannot be represented.
BTW, the clipping detection algorithm is not very intelligent and hence misses the clipping in the recording the moment you turn the level down ever so slightly.

In a real DAC chip, depending on how much headroom there is for IS-overs and what exactly happens when headroom runs out, there might zero issues (no clipping or a touch of soft-clipping) but also catastrophic ones (wraparound etc).

--------:-------

We must remember that IS-overs themselves are not bad and not a fault. Badly handled IS-overs of reasonably limited strength (say, IS-over that would come out below 2dB even with the most sinc'ish filter if properly handled) are the problem.

 

[melowman]

“a properly done master”?
If Audacity bases its analysis only on samples being at full scale, then any master peaking below 0dBFS (-0.1dBFS for example, or whatever) will show up like your picture below!

Did another test. This time with a properly done master.

Now THIS is conclusive evidence:

View attachment 405355

0 clipped samples. Tons of ISOs. Clipping here is entirely created by oversampling.

In this case, the probability of having ISOs without any clipped samples is 100%, wouldn’t you agree?

 

[tccalvin]

In a real DAC chip, depending on how much headroom there is for IS-overs and what exactly happens when headroom runs out, there might zero issues (no clipping or a touch of soft-clipping) but also catastrophic ones (wraparound etc).

Hence the need for testing. Now hopefully we can get back on topic lol.

Does anyone have any ideas about potential tests we could do besides the ones that have already been proposed? Or maybe some criticisms or ways to improve the existing methods?

 

[tccalvin]

“a properly done master”?

Well… I meant a master that doesn’t have any clipped samples above 0dBFS. Whether that’s a “proper master” or not is a matter of opinion, I guess.

If Audacity bases its analysis only on samples being at full scale, then any master peaking below 0dBFS (-0.1dBFS for example, or whatever) will show up like your picture below!

Yes…? Isn’t that the whole point of the test? That’s the kind of music we’re dealing with these days…

 

[melowman]

Take any crushed to death, awful sounding master, and lower it just 0.1dB (or leave as is if it’s peaking below 0dBFS) and it will show up like in your picture, ‘red-free’.

 

[restorer-john]

And you still don't seem to understand what inter-sample overs are.

Clearly you are the one who doesn't get it.

you can have ton's of IS-overs in otherwise well-recorded material.

No. If you have "tons of inter-sample overs" the material is NOT well recorded by definition.

And if you are unnecessarily oversampling already too hot material, I have no sympathy for the mess that ensues.

The entire "issue" is a solution in search of a problem caused by the inability to record properly in the first place. You will have no intersample overs, ever, if you record with plenty of headroom and stop upsampling for no good reason. Not hard. We've been doing it for decades. My first DAT (1990) had an "over" indicator. What do you think that was set for? I'll let you guess and it wasn't 0dBFS.

 

[tccalvin]

Take any crushed to death, awful sounding master, and lower it just 0.1dB (or leave as is if it’s peaking below 0dBFS) and it will show up like in your picture, ‘red-free’.

Right. We should specify that you're talking about what's been done at the mastering stage. I haven't touched the levels at all.

I also personally want dynamics in music. It's just a hard sell for people. Everyone can see the difference between SDR and HDR in video, but you need trained ears to understand dynamic range in music.

There is also probably something to be said about creative intent and wanted vs unwanted distortion. A mastering engineer might have struck a compromise between loudness and audible distortion when listening to the finished product in the studio. However, listening to the CD master through an oversampling DAC with no headroom might make the signal even more distorted than it originally was.

Look, I get we don't live in the world we want to live in. It is what it is. Loud music sells more, or at least that is what is believed by the music industry. And them believing it is all it takes to make it a reality. The question is, to what extent can we mitigate this issue once the damage has been done?

 

[Sokel]

Right. We should specify that you're talking about what's been done at the mastering stage. I haven't touched the levels at all.

I also personally want dynamics in music. It's just a hard sell for people. Everyone can see the difference between SDR and HDR in video, but you need trained ears to understand dynamic range in music.

There is also probably something to be said about creative intent and wanted vs unwanted distortion. A mastering engineer might have struck a compromise between loudness and audible distortion when listening to the finished product in the studio. However, listening to the CD master through an oversampling DAC with no headroom might make the signal even more distorted than it originally was.

Look, I get we don't live in the world we want to live in. It is what it is. Loud music sells more, or at least that is what is believed by the music industry. And them believing it is all it takes to make it a reality. The question is, to what extent can we mitigate this issue once the damage has been done?

I would think of clients too about their music,not only the engineers.
Presenting them the outcome (any outcome) with studio's mains monitors are shock and awe,specially the younger ones involved with metal-ish and pop stuff.
Specially for the first time.

Anything is insignificant at the face of an experience like that.Even audiophiles not used in big sound would be impressed.

So...