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

[danadam]

His reasoning is, that for it to happen, the signal before downsampling must have clipped. This clipping gets picked up at recording/mixing, and any engineer worth his salt, would have taken care of it, before downsampling.

And to you this is different from "he says intersample-overs can only happen during recording (or downsampling)"? Ok. To me it's the same, only using different words.

 

[Ken Tajalli]

And to you this is different from "he says intersample-overs can only happen during recording (or downsampling)"? Ok. To me it's the same, only using different words.

my bad

 

[KSTR]

Intersample-Overs can happen anytime a signal is processed (notably limiting and soft-clipping), resampled, upsampled (like in the DAC itself during playback), etc. Even with a perfect natural recording to start with. Intersample-Overs are a fully legit thing and they only pose problems when they are handled in bad way (simple soft-clipping is benign, and that's what most DAC chips do once IS-overs are larger than a few dB).
And yes, IS-Overs can be fully avoided by proper pre-processing at the mastering stage but it takes knowledge to do it properly, plus the right tools.

 

[morillon]

These are situations that can easily occur, if not compensated, during "acoustic" correction.

 

[edechamps]

But you came to the conclusion that it's an Android software issue, with DAC chip proper innocent, didn't you?

No, it's the opposite. The worst cases are actually the ones where Android is NOT resampling and the data is being fed directly to the DAC. It's clearly the DAC that's doing the "slamming". (This makes sense because, if you take a test signal with deliberate intersample overs and pass it through a resampler, it's likely the resampling operation will end up clamping the overs somewhat, just because the samples are shuffled around in time.)

 

[VientoB]

Read it again, that's not what he is saying.
He is saying intersample overs hardly ever occur. His reasoning is, that for it to happen, the signal before downsampling must have clipped. This clipping gets picked up at recording/mixing, and any engineer worth his salt, would have taken care of it, before downsampling.
At least, that is my understanding.

But how about if it doesnt clip in the analogue, and then is adjusted in the digital to maximise the signal? Surely that must be what is being done otherwise people would not be finding inter-sample overs in their analysis?

 

[Ken Tajalli]

But how about if it doesnt clip in the analogue, and then is adjusted in the digital to maximise the signal? Surely that must be what is being done otherwise people would not be finding inter-sample overs in their analysis?

Well I am no expert, but who has found IS-overs in their analysis? it is so rare.
Recordings are done in hires. If pre downsampling there are no overs , how can it appear afterwards?
I can think of one scenario, if you upscale a near zero peak material, and you be unlucky that the IS falls in the wrong place. Even that would be for a very short time. So if it is going to be an oversampling, allow for the oversample, as Chord does.
My 2 pence, though again somebody should correct me.

 

[markstein]

But how about if it doesnt clip in the analogue, and then is adjusted in the digital to maximise the signal? Surely that must be what is being done otherwise people would not be finding inter-sample overs in their analysis?

That is exactly what is most probably happening.

The last step in mastering - volume adjustment - does not seem to be done correctly in many cases. For no objective reason, other than stupid loudness maximization.

 

[markstein]

Well I am no expert, but who has found IS-overs in their analysis? it is so rare.
Recordings are done in hires. If pre downsampling there are no overs , how can it appear afterwards?
I can think of one scenario, if you upscale a near zero peak material, and you be unlucky that the IS falls in the wrong place. Even that would be for a very short time. So if it is going to be an oversampling, allow for the oversample, as Chord does.
My 2 pence, though again somebody should correct me.

It is not rare.

Scanning my library of digitized CDs with jRiver MC library tools shows so many true peaks above 0dBFS that I would not call it a seldom phenomenon.

Even when streaming with Tidal lossless, my RME ADI-2 true peak meter shows overs with some songs...

 

[markstein]

Btw. this is the reason why I sold my minidsp SHD. The inbuilt ASRC has no means of attenuating the input signal, so intersample overs cause clipping during the upsampling to 96kHz.
I have activated the optional 3dB attenuation prior to SRC in my RME ADI-2, and in my opinion this should be a standard feature.

Edit:
Funny side story: I contacted the company Mutec offering "professional SRC devices" like the MC4 (very famous in German audiophile.circles some time ago) and asked them whether the input signal could be attenuated to prevent intersample overs. They were not even able to answer this question...

 

[markstein]

Maybe @mdsimon2 can contribute to this topic as well. I learned a lot from his posts regarding implementation of digital eq filters.

See also this post:

DSP Measurements and Rising Noise Floor

If you scan a library of lossless CD tracks with an upsampling reconstruction filter, you will find that most overs are less than +2 dBFS. I haven't seen any over +3 dBFS, and I'm not entirely sure that anything above 3.01 dBFS is possible when the input signal has been properly band limited. In...

audiosciencereview.com

 

[VientoB]

Well I am no expert, but who has found IS-overs in their analysis? it is so rare.

From what I've seen it's not rare.

 

[John_Siau]

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.

The assumption that "a huge amount of music ... doesn't have this problem" is incorrect.

The vast majority of 44.1 kHz recordings have intersample peaks that exceed 0 dBFS. Curiously, tracks with minimal compression can have some of the highest and most frequent intersample peaks. See my application note on this topic: Intersample Overs in CD Recordings

When intersample peaks overload a DAC, a DSP, or an ASRC, the distortion reaches very high levels and it is often non-harmonic. This playback defect is huge compared to all of the other THD, IMD, and SNR issues in playback hardware.

I would assert that the intersample over test is the single most important test from the standpoint of detecting audible defects in a digital audio product.

The cure requires a 3-dB loss in SNR. This a small price to pay for eliminating the single most audible defect with most DACs, especially when most modern DACs have a higher SNR than the power amplifiers that they will be driving.

 

[John_Siau]

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.

If you were testing cars, would you just test for horse power and torque but not actually drive the car above 60 MPH? If something really bad happens at 70 MPH, your readers would want to know about it. This is especially true when all of your readers would reasonably be expected to drive at those speeds.

DAC users will encounter intersample peaks above 0 dBFS and may encounter them several times per second, even on well-recorded tracks.

Do the wheels fall off at 70 MPH? Your readers need to know!

 

[KSTR]

The assumption that "a huge amount of music ... doesn't have this problem" is incorrect.

The vast majority of 44.1 kHz recordings have intersample peaks that exceed 0 dBFS. Curiously, tracks with minimal compression can have some of the highest and most frequent intersample peaks. See my application note on this topic: Intersample Overs in CD Recordings:

Every D/A chip and SRC chip that we have tested here at Benchmark has an intersample clipping problem! To the best of our knowledge, no chip manufacturer has adequately addressed this problem.

(insert mine).
Well, IME there is one DAC chip that really does quite well with IS-Overs, the AK4493 (first version, haven't checked the new post-fire models).
Up to 3dB Overs are handled with minimal deformation (depending on filter choice) and above 3dB Overs are nicely soft-clipped. Of course the analog output circuitry in a specific DAC model must be able to cope with that extra voltage without clipping.

For most other chips, including any ESS models, digital reduction is the simple fix, just like you say (and I would agree that the SNR penalty is negligible).

 

[AnalogSteph]

Just for the record compact disc, here's some 0dBFS+ test data for misc. CD players taken by @NTTY, Nielsen/Lund style:

Teac VRDS-20 Review (CD Player)

Nice report, test and an even nicer device. I also own a Teac VRDS-25x, an absolute tank among CD players. I still need a VRDS-20 for my pleasure, if only because of the look and construction. Not even one of the Teak developers could explain to me at an event what Teak was thinking with the...

www.audiosciencereview.com

 

[John_Siau]

(insert mine).
Well, IME there is one DAC chip that really does quite well with IS-Overs, the AK4493 (first version, haven't checked the new post-fire models).
Up to 3dB Overs are handled with minimal deformation (depending on filter choice) and above 3dB Overs are nicely soft-clipped. Of course the analog output circuitry in a specific DAC model must be able to cope with that extra voltage without clipping.

For most other chips, including any ESS models, digital reduction is the simple fix, just like you say (and I would agree that the SNR penalty is negligible).

I believe the AK4493 places the digital volume control in front of the interpolator. The ESS places it after the interpolator. Turning the digital volume down inside the ESS does not add headroom. The digital level must be turned down before sending the signal to the ESS.

It must also be turned down before any SRC or ASRC process and before many DSP processes. A digital attenuation of 3 dB is sufficient for just about any source material.

 

[miero]

At least datasheet of the ES9039PRO shows volume control before the OSF & ASRC. And a similar diagram is also in the ES9039Q2M datasheet.

So regarding the $SUBJ it might be a good idea to choose a DAC with newer ESS chip (Hyperstream IV) if DAC manufacturer does not attenuate signal in another chip DSP/XMOS/FPGA/...

Or to attenuate signal in the player if possible.

 

[restorer-john]

I have yet to see proof that intersample overs are an actual, audible issue.

Take a pure 1kHz sine. At +3dB overload, the waveform and samples look like this:

It's very audible.

+3dB overload on complex music you likely won't hear. I've got DAC overload test CD tracks (music) where +9dB is audible, but certainly not unlistenable.

 

[amirm]

When intersample peaks overload a DAC, a DSP, or an ASRC, the distortion reaches very high levels and it is often non-harmonic. This playback defect is huge compared to all of the other THD, IMD, and SNR issues in playback hardware.

That overload happens in a fraction of a second in some music whereas the distortion products you mention are constant. There is a reason the world is not up in arms about this.

As you know, the fix for this is not free. We lose dynamic range for all content and all times. The real solution would be to avoid this in production of the music, not consumption. Or use DSP volume control upstream of the DAC.