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

[melowman]

I’m not exactly sure how you came up with the claim that digital hard clipping isn’t audible. Digital hard clipping is audible!

Also, loudness and the amplitude of inter-sample levels aren’t necessarily related; there’s plenty of ‘reasonably-loud’ material showing +2/3dB inter-sample values. High inter-sample values are more connected to high high-frequency transients — which loud music may have, but so does reasonably-loud music too. And as you stated, DAC hard clipping in loud material may be less noticeable than with softer material. In other words, hard clipping may get more noticed with softer material.

So this is an additional argument to handle inter-sample hard clipping in DAC designs.

 

[KSTR]

I’m not exactly sure how you came up with the claim that digital hard clipping isn’t audible. Digital hard clipping is audible!

That's not what I said. I stated that clipped IS-overs are not audible. The reason is the required specific signal content for them to appear in the first place.
If in doubt, try ABX'ing clipped IS-overs vs. unclipped ones, see my post #64

 

[melowman]

IS overs are hard clipped in oversampled DACs. The clipping event might (or might not) be shorter for an oversampled signal, but still, it is hard clipping.

(Btw in the test you mention, no hard clipping would occur, since all inter-sample values would be under 0 after applying the -6dB gain in your protocol. But the oversample filter might overshoot, though, which leads me to that question: are you considering that clipping in OSed DACs are only due to the OS filter?)

 

[KSTR]

Btw in the test you mention, no hard clipping would occur, since all inter-sample values would be under 0 after applying the -6dB gain in your protocol

Please read more carefully. The hard-clipping is applied by signal processing to properly emulate hard-clipped IS-overs, no DAC involved at this point. And the scaling (and final 4x output rate) is necessary to avoid the DAC used for playback entering the picture, obviously.

 

[KSTR]

are you considering that clipping in OSed DACs are only due to the OS filter?

There are several places where the clipping can occur
- in the digital filter
- in the delta-sigma modulator
- in the analog output stage (including on-chip amps of voltage-output DAC chips)

 

[edechamps]

By the way for those willing to experiment, here are some SoX commands to generate ISO test signals at various true peak levels at any given sample rate and bit depth:

sox --channels 2 --bits 16 --rate 48000 --null intersample-peaks-48000-16-0.wav synth 30 sine 12000 0 12.5 gain -1
sox --channels 2 --bits 16 --rate 48000 --null intersample-peaks-48000-16-1.wav synth 30 sine 12000 0 12.5 gain 0
sox --channels 2 --bits 16 --rate 48000 --null intersample-peaks-48000-16-2.wav synth 30 sine 12000 0 12.5 gain 1
sox --channels 2 --bits 16 --rate 48000 --null intersample-peaks-48000-16-3.wav synth 30 sine 12000 0 12.5 gain 2
sox --channels 2 --bits 16 --rate 48000 --null intersample-peaks-48000-16-4.wav synth 30 sine 12000 0 12.5 gain 3

The generated signals look like this:

Worth noting that the period is 12 kHz so well below Nyquist.

And this is what bs1770gain has to say about them:

  [1/5] "intersample-peaks-48000-16-0.wav":
       true peak:  -0.58 TPFS / 0.935585
  [2/5] "intersample-peaks-48000-16-1.wav":
       true peak:  0.42 TPFS / 1.049707
  [3/5] "intersample-peaks-48000-16-2.wav":
       true peak:  1.42 TPFS / 1.177829
  [4/5] "intersample-peaks-48000-16-3.wav":
       true peak:  2.42 TPFS / 1.321535
  [5/5] "intersample-peaks-48000-16-4.wav":
       true peak:  3.42 TPFS / 1.482774

Some time ago I did ISO measurements on the DAC bundled with a Google Pixel 3, a Moshi DAC, and a cheap chinese VIMVIP DAC. Some of these measurements show waveform distortion that is significantly worse than what would be caused by hard clipping alone. At the highest TP level listed above, all devices I tested exhibit extreme non-linear distortion all over the spectrum that is not even harmonically related to the original 12 kHz tone - a phenomenon that cannot be explained by clipping alone.

 

[amirm]

If the solution is to reduce dynamic range, it means that you disadvantage huge amount of music that doesn't have this problem in order to deal with small amount of content. In other words, if the headroom is permanently built into the DAC, then you lose that dynamic range for all uses, not just in the case of intersample overs. This is one of the reason that the DACs that have this feature are not at the top of our chart.

The right solution is to have an option in the music player to perform the dynamic range reduction on a per track basis. An online database could be built to instruct the player to do this automatically.

 

[amirm]

BTW, I have had the Benchmark test for 2 to 3 years now. The reason I have not deployed it is above. Us forcing it will mean all DACs will eventually lose 3 dB of dynamic range or whatever this needs to be. So best think about what you are asking. There are plenty of tracks I avoid because fidelity is so poor. I don't mind having tracks with audible issues due to clipping added to that list.

 

[melowman]

But @amirm you’re the one saying best case scenario is a threshold of hearing at -115dB…. So what’s the deal?

 

[amirm]

But @amirm you’re the one saying best case scenario is a threshold of hearing at -115dB…. So what’s the deal?

Meaning what? You need to EQ your system. You need headroom for that. If you deploy codecs, they need headroom for that. If you perform speaker correction separate from room EQ, you need headroom for that. This is why AVRs have such poor dynamic range. Engineers work hard to give us the best dynamic range possible. You have to be careful in saying you want to throw away some of it permanently for 0.001% of content out there with this issue.

And remember, as stated, you always have the option of buying a DAC with volume control. So it is not like you don't have a solution like that.

 

[melowman]

@amirm that’s certainly not that very tiny percentage, this is pretty exaggerated.

You are correct that whomever applies DSP, needs to make sure there’s enough headroom. In that regard, it is OSed DAC manufacturers who apply DSP (oversampling), so by your rule they need to make room in order their DSP to no clip.
This goes by your logic. And I agree with it.

 

[amirm]

One more thing: I already test for 0 dBFS. Stereophile doesn't and uses lower levels. Just last week a manufacturer complained that at 0 dBFS they have problems and they wanted me to test at lower levels. In other words, I am already pushing the industry to produce clean signal when the source is at maximum PCM value. What is being asked here is to push even more. We have to be careful about that.

 

[melowman]

Please read more carefully. The hard-clipping is applied by signal processing to properly emulate hard-clipped IS-overs, no DAC involved at this point. And the scaling (and final 4x output rate) is necessary to avoid the DAC used for playback entering the picture, obviously.

I did read. And I read that the initial track has inter-samples below +6dBFS. Scaling that by 0.5 will give inter-samples below 0dBFS, so no hard clip occurs. Unless you meant that the applied hard-clipping should occur at a -6dBFS threshold, in which case I misunderstood your intention and approach.

But that was not the point of my answers to you. I’m working with digital hard clippers quite often and I carefully choose my threshold for best effect: a few dBs of hard clipping are definitely audible.

And as reminded, the less crushed the sound is in the first place, the more likely hard clipping will heard.

All of that obviously depends on the material.

 

[melowman]

One more thing: I already test for 0 dBFS. Stereophile doesn't and uses lower levels. Just last week a manufacturer complained that at 0 dBFS they have problems and they wanted me to test at lower levels. In other words, I am already pushing the industry to produce clean signal when the source is at maximum PCM value. What is being asked here is to push even more. We have to be careful about that.

I don’t get why the hand is trembling that much. Manufacturers used to be afraid by you, but now the situation seems to be have been inverted. No offense or lack of respect here. I’m not understanding your last two sentence..

SNR seems to have become an irrational golden end goal making people blind to other important issues.
Yes manufacturers have been hard at trying to keep dynamic range as high as possible, but at the expense of other obvious issues.

Past a point, dynamic range is just a commercial argument, we’re not hearing that stuff (you’re the first, Amir, to brandish the argument of threshold of hearing), and I think that reviews in a website called audio science review shall not transpire that (I’m not saying this is the intention here). I think scientific issues should be addressed, hence the OP.

You were first, Amir, to fight for clean signal reproduction, and you made things move amongst manufacturers. The quest is not over yet, there are still issues to address! Your voice counts!

 

[Adis]

I don’t get why the hand is trembling that much. Manufacturers used to be afraid by you, but now the situation seems to be have been inverted. No offense or lack of respect here. I’m not understanding your last two sentence..

SNR seems to have become an irrational golden end goal making people blind to other important issues.
Yes manufacturers have been hard at trying to keep dynamic range as high as possible, but at the expense of other obvious issues.

Past a point, dynamic range is just a commercial argument, we’re not hearing that stuff (you’re the first, Amir, to brandish the argument of threshold of hearing), and I think that reviews in a website called audio science review shall not transpire that (I’m not saying this is the intention here). I think scientific issues should be addressed, hence the OP.

You were first, Amir, to fight for clean signal reproduction, and you made things move amongst manufacturers. The quest is not over yet, there are still issues to address! Your voice counts!

If I may: fighting for clean signal reproduction ≠ fighting for not ruining program material by clipping it. Once there were loudness wars. And loudness won.

 

[KSTR]

Unless you meant that the applied hard-clipping should occur at a -6dBFS threshold, in which case I misunderstood your intention and approach.

Yes, that's what I meant.

 

[melowman]

If I may: fighting for clean signal reproduction ≠ fighting for not ruining program material by clipping it. Once there were loudness wars. And loudness won.

We’re talking clipping in DACs here… ! : )

Yes you’re right: let not music equipment « ruin » audio material by adding unwanted (and avoidable) distortion that’s not here in the first place.

On the other side, ‘artistic’ clipping has been an artistic tool for decades. Be it from artists or mixing and mastering engineers. Rock music wouldn’t exist without clipping, and the list goes on, there’s ‘artistic’ clipping baked in a gigantic amount of pieces of music, including the ones you listen to.

But ‘artistic’ clipping ain’t faulty technical clipping — as Amir reminded, if you apply DSP, you must make sure there is enough room for the result! Not making room is a design mistake!

 

[Adis]

We’re talking clipping in DACs here… ! : )

Yes you’re right: let not music equipment « ruin » audio material by adding unwanted (and avoidable) distortion that’s not here in the first place.

On the other side, ‘artistic’ clipping has been an artistic tool for decades. Be it from artists or mixing and mastering engineers. Rock music wouldn’t exist without clipping, and the list goes on, there’s ‘artistic’ clipping baked in a gigantic amount of pieces of music, including the ones you listen to.

But ‘artistic’ clipping ain’t faulty technical clipping — as Amir reminded, if you apply DSP, you must make sure there is enough room for the result! Not making room is a design mistake!

Soooo, like, sanction the consequences, not the causes?

 

[KSTR]

But that was not the point of my answers to you. I’m working with digital hard clippers quite often and I carefully choose my threshold for best effect: a few dBs of hard clipping are definitely audible.

In general I would certainly agree. Chopping off a relatively large number of samples in a row of a tonal signal at low/medium frequencies (say, a mellow kick drum, plucked upright bass, nylon string guitar, piano, stuff like that, not to forget vocals) is quite audible.
The question is whether a single clipped intersample over -- which cannot be any longer than one sample period -- is audible. A whole train of such events in a specific test signal (like the fs/4 sine -- or enveloped fs/4 sine burst -- with 45degree start offset, fully normalized) might have some chances of becoming audible, but isn't that some unrealistic corner case that will never happen with real music signals?

And as reminded, the less crushed the sound is in the first place, the more likely hard clipping will heard.

All of that obviously depends on the material.

It would be great if you could provide a song snippet with a "low crush factor" where you think clipped vs. unclipped intersample overs would be audible so we could set up a blind test?

 

[melowman]

I can try to provide something publicly shareable, and my own ABX test results. (Don't hold your breath for it to happen right now, though. )