Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Dolby jumps on the upsampling bandwagon (Read 12437 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Dolby jumps on the upsampling bandwagon

apparently it's not just 44.1 kHz that's inadequate, it's 48khz too

http://www.dolby.com/uploadedFiles/Assets/...ormance.PDF.pdf

whistled version:

http://www.stereophile.com/content/dolby17...apodizes-truehd


The technical gist is that Dolby's using Meridian's 96kHz upsampler and 'apodizing' filter, which they say masks pre-ringing.
(which apparently is commonly audible using 48kHz SR, and apparently can be masked without introducing frequency response or phase
artifacts that are at least as audible -- right? Because otherwise there would be no point to this...right? )

Btw, my respectable but not-boutique-high-end AVR already upsamples everything to 96kHz (in order to apply DSP) so again, WTF, Dolby?

Dolby jumps on the upsampling bandwagon

Reply #1
They're promising to eliminate preringing on the production end.

Btw, my respectable but not-boutique-high-end AVR already upsamples everything to 96kHz (in order to apply DSP) so again, WTF, Dolby?

Is there only one way to upsample?

Honestly, is it being suggested that preringing is never audible with any steep filter, no matter the implementation or how many times it is used in the production chain?

Dolby jumps on the upsampling bandwagon

Reply #2
They're promising to eliminate preringing on the production end.

Btw, my respectable but not-boutique-high-end AVR already upsamples everything to 96kHz (in order to apply DSP) so again, WTF, Dolby?

Is there only one way to upsample?

Honestly, is it being suggested that preringing is never audible with any steep filter, no matter the implementation or how many times it is used in the production chain?


Of course not. As usual, the existence of a large number of "good" reconstruction filters that are sonically innocent is being suggested if not appealed to.

Dolby jumps on the upsampling bandwagon

Reply #3
They're promising to eliminate preringing on the production end.

 
There are tradeoffs to that, of course.  And it's not the only promise they're making, of course.

Quote
Is there only one way to upsample?


No, but are the ways used in mass market gear often audibly deficient?

Quote
Honestly, is it being suggested that preringing is never audible with any steep filter, no matter the implementation or how many times it is used in the production chain?


No.  It is being suggested, by Dolby Labs with the help of Meridian (which has never shirked from proposing 'solutions' to problems of dubious audible import) , that 1) pre-ringing is both audible and common in 48kHz audio, and 2) common upsamplers (and upsampling has been a feature of consumer DACs for , what, 20 years now?) are audibly distinguishable from expensive ones and 3) the result of their special upsampling/filtering audibly improves the original signal (I believe the term 'restores the natural tonality' has been used in some of their copy). 

Proof for any of these suggestions?  Not provided.  Non-blind demos to 'critics' flown in from around the country?  They're on that.











Dolby jumps on the upsampling bandwagon

Reply #4
And it's not the only promise they're making, of course.
Of course? Tell me, what other promises are they making? It seems you didn't really read the paper you linked.

Quote
No, but are the ways used in mass market gear often audibly deficient?
Non-sequitur. They made a point of saying there is no need for the end-user to buy new equipment. Again, did you actually read the paper?

Quote
the result of their special upsampling/filtering audibly improves the original signal (I believe the term 'restores the natural tonality' has been used in some of their copy).
Do you think audible pre-ringing is part of natural tonality?

Quote
Proof for any of these suggestions?  Not provided.
Do you think that they don't employ double-blind testing during development?  Have you ever visited their facilities?

Dolby jumps on the upsampling bandwagon

Reply #5
Of course? Tell me, what other promises are they making? It seems you didn't really read the paper you linked.


It seems you missed that there were two links.  The second one has a slide from Dolby's presentation to the press, where it claims several things rather specifically to be audible improvements:



And you are also being rather pedantic, since *of course* there's more than just those two sources informing me. There's this press release I alluded to further down in my post, for example.  It's where the 'natural tonality' ref came from.  Furthermore  Dolby's new claims are being discussed on at least one high-level, engineering-heavy pro audio group I'm aware of -- it's where I first heard of them, in fact.  Fair use quote from a participant there: "They just have a nice sounding filter in their upsampler, and are trying to convince people that somehow it is more "natural" than just playing back the 48k material, or upsampling with a proper linear phase filter." 

(versus Meridian's 'apodizing' minimum phase filter, that is)

Quote
Quote
No, but are the ways used in mass market gear often audibly deficient?
Non-sequitur. They made a point of saying there is no need for the end-user to buy new equipment. Again, did you actually read the paper?


Yes, I actually did.  And no, not a non-sequitur.  Yes, Dolby is coding their *audible* improvement of the 48kHz source right into the signal, by passing it through their high-quality 96kHz upsampling + pre-ring-masking filtering, prior to recoding.  So no new hardware needed.  Whereas by implication, the 96kHz upsampling and filtering my AVR does by itself will have no such salutary audible effect, and may even make things worse.

This is surely not impossible, but it's not demonstrated by Dolby.


Quote
Do you think audible pre-ringing is part of natural tonality?


No, but how do they, or I, know that the output of their filter has no effect on 'the natural tonality' either?

Quote
Do you think that they don't employ double-blind testing during development?  Have you ever visited their facilities?


No.  Nor have I seen them mention DBT for any this either.  That's (part of) the problem.    I'd like to see that there was such testing. And what the results were, if only to get  a sense of how common this 'problem' is.

I'm holding Dolby claims to HA standards.  So you can stop attributing strawman argument to me now, thanks.

Dolby jumps on the upsampling bandwagon

Reply #6
The second one has a slide from Dolby's presentation to the press, where it claims several things rather specifically to be audible improvements:
So you don't believe elimination of audible preringing could deliver the benefits listed?

Quote
Whereas by implication, the 96kHz upsampling and filtering my AVR does by itself will have no such salutary audible effect, and may even make things worse.
There is no such implication.

Quote
how do they, or I, know that the output of their filter has no effect on 'the natural tonality' either?
They've have provided some basis for potential improvement. How about you? All I'm reading here is you're fearful that they might have screwed something up.

Quote
I'm holding Dolby claims to HA standards.
Nothng wrong with that. So now what, we speculate and hope that someone from Dolby chimes in?

Dolby jumps on the upsampling bandwagon

Reply #7
The second one has a slide from Dolby's presentation to the press, where it claims several things rather specifically to be audible improvements:
So you don't believe elimination of audible preringing could deliver the benefits listed?


Obviously, elimination of audible preringing could have some benefits. The audible benefits would follow the degree to which this alleged problem exists, if it exists at all.

The Dolby article (at least implicitly) claimed that many, perhaps the vast majority of inexpensive Blu ray players have this fault.

I own or otherwise use several inexpensive Blu ray players that I listen to frequently. I seriously doubt that they have this fault.

It is easy to forecast that should someone actually do real world testing, Meyer and Moran's JAES article will have a logical sequel. ;-)

Dolby jumps on the upsampling bandwagon

Reply #8


No, but are the ways used in mass market gear often audibly deficient?



This question is of the essence.

Quote
Non-sequitur. They made a point of saying there is no need for the end-user to buy new equipment. Again, did you actually read the paper?


With all due respect, the above question contains enough non-sequiturs to choke a horse. ;-)

The first non-sequitur is dismissing the question that is the crux of the matter. Is there even a problem to solve?

The issue of end users buying new equipment is a non sequitur because whether or not their existing equipment has an audible fault due to pre-ringing is the very crux of this matter.

If the existing consumer equipment has this audible flaw, then there can be a discussion of solutions for it.

If the flaw does not exist, then discussion of solutions for it are out of order on the grounds of relevance.

Providing proof that we have read the paper is made difficult by the fact that it is copy protected in such a way that many of us can't cut and paste from it.

Dolby jumps on the upsampling bandwagon

Reply #9
Do you think that they don't employ double-blind testing during development?  Have you ever visited their facilities?


There is anecdotal evidence that Dolby has done DBTs as part of their development projects in the past.

However, they can change their policies and practices without notice, right?

I speculate that the need for in-house DBTs was finessed by the perpetrators of this project on the grounds that they bought the technology from Meridian.

Unfortunately, Meridian's stance on their use of DBTs seems vacant.

Dolby jumps on the upsampling bandwagon

Reply #10
Never mind the audibility of the "problem" they're solving - what does it say about the world when British company Meridian has to licence their British inventions through Dolby to get the world to notice them.

First MLP, now apodizing filters.

If Michael Gerzon had understood that the world works in this way, we could have been listening to Dolby Ambisonics for the last 30 years.

Cheers,
David.

(I don't know if this post deserves / needs a smiley, or not)

Dolby jumps on the upsampling bandwagon

Reply #11
So you don't believe elimination of audible preringing could deliver the benefits listed?


I don't believe Dolby has demonstrated it delivers *those* benefits.

Quote
There is no such implication.


Pedantry again.  So, if not to claim an audible improvement over the status quo, why does this product exist then?  96kHz upsampling and anti-imaging filters (including Meridian's) aren't new tech

Quote
They've have provided some basis for potential improvement. How about you? All I'm reading here is you're fearful that they might have screwed something up.


And I'm reading an extraordinary defensiveness re: Dolby -- a former employer or what?  Now I'm curious.  FWIW I have no animus against them -- even defended them against both the 'DTS uses higher bitrates therefore better' crowd and the 'lossy is just awful compared to lossless' crowds.  And I'm a a huge fan of their DPL IIX product.


Quote
Nothng wrong with that.


Glad to hear it.

Quote
So now what, we speculate and hope that someone from Dolby chimes in?



Of course not, because on HA we always sit quiet and wait for the manufacturer to come over and explain the basis of their claims.

 

Personally, I'm going to 1) keep reading what Dolby is saying  and 2) keep checking in on what people who actually design ADCs and DACs are saying about this.  I'd also expected that the technical expertise of curious HA members might
be one of the 'chimes'. 

What else do you suggest?  Shall we petition Dolby to provide the 48kHz sources they used, plus the processed versions, so we can do the DBTs they so far haven't reported?

Dolby jumps on the upsampling bandwagon

Reply #12
The Dolby article (at least implicitly) claimed that many, perhaps the vast majority of inexpensive Blu ray players have this fault.

There was no such claim, nor was there any such implication. Let's have a look at what the paper actually says instead of conviently pretending it says something different, m'kay?!? (same goes for krab)
Quote
Movie soundtracks are constructed from a wide variety of audio elements—dialogue, music, special effects—that are mixed together in the final production. Preringing from A-to-D conversion of these sources can be introduced any time in the content-creation process.

[...]

Dolby’s approach to solving this problem is to fully implement Meridian’s proven upsampling process and to shift the computational burden of this processing to Dolby Media Producer,
the professional product that encodes Dolby TrueHD bitstreams for Blu-ray Disc. 

[...]

Content encoded with Dolby TrueHD with advanced 96k upsampling is fully playback compatible with all Blu-ray disc players and A/V receiver products. No new hardware or upgrades are needed to experience the benefits of this technology.


Providing proof that we have read the paper is made difficult by the fact that it is copy protected in such a way that many of us can't cut and paste from it.

What are you on about? I just copied and pasted from it.

Dolby jumps on the upsampling bandwagon

Reply #13
When we realize this pre-ringing is happening at frequencies we normaly can´t hear, so above or around the cut-off frequency of the signal we should wonder why we should be able to hear it.
I was told there were some older DAC designs that had a problems handling this and must have created noise below that became audible. Todays DACs should have no problems handling this and cut below the frequencies that ringing happens most anyway.

Benchmark Media claims this is even a non-issue and is convinced it does more harm as good but they talk about image fold back. I doubt the Meridian filter does have much of it. Even the non-linearity of the apodizing filter should harm more as any pre-ringing.
You can read here:
http://www.benchmarkmedia.com/discuss/foru...odizing-filters
Of course someone might think they have to claim such things because they don´t have the patent to do otherwise

Is troll-adiposity coming from feederism?
With 24bit music you can listen to silence much louder!

Dolby jumps on the upsampling bandwagon

Reply #14
Providing proof that we have read the paper is made difficult by the fact that it is copy protected in such a way that many of us can't cut and paste from it.

What are you on about? I just copied and pasted from it.


Do tell. Marking and copying to the clipboard doesn't happen with Adobe Reader X or the Acrobat reader in Google Chrome. Open Office Draw won't open it at all, claiming that it is encrypted.

Dolby jumps on the upsampling bandwagon

Reply #15
Using an iPad.  Thankfully I'm not being accused of manufacturing a fictitious quote, but I guess there's still time for that. 

Dolby jumps on the upsampling bandwagon

Reply #16
This press release provides some details:
http://investor.dolby.com/ReleaseDetail.cf...eleaseid=674604

Dolby suggests that the majority of professional ADC's suffers from pre-ringing. I have no problem with that since they all use linear-phase anti-aliasing filters. They are a bit vague about its audibility:
"the detrimental effect of preringing introduced during analog-to-digital conversion of the recorded signal."
and
"These digital artifacts can introduce an unnatural edginess or harshness to the audio. Dolby TrueHD with advanced 96k upsampling restores the natural tonality of the soundtrack."

From the Pro-Audio mailinglist krabapple was referring to I understand that the non-linear-phase apodizing filter modifies the phase response of the output, very likely also in the audible band under 20 kHz. While some people might judge this as an improvement in sound quality, the only way to be sure is to compare (ABX) with a "true-hi-res" version in 24/96 or even 24/192 format. It was suggested on the Pro-Audio list that the phase behaviour of the filter might have a euphonic effect, but in that case it should be called an effect and not a correction of a presumed defective ADC.

What surprises me is that the upsampled content content will be played on 96 kHz DACs with pre-ringing linear-phase filters. Apparently Dolby considers pre-ringing at 96 kHz less of a problem.

Dolby jumps on the upsampling bandwagon

Reply #17
Obviously, elimination of audible preringing could have some benefits. The audible benefits would follow the degree to which this alleged problem exists, if it exists at all.

Not end-user players and receivers, this is the "essence" of the debate (obnoxious font size not quoted).  Let there be no mistaking, like the rest of you I am highly skeptical that this technology will actually provide any tangible benefit (that is also sonically transparent!) for the very reason that Wombat gave.  The problem is that we don't really get very far when we demonstrate woeful reading comprehension with such belligerence.

...and yes, the word "proven" in the excerpt I provided makes me cringe.

Dolby jumps on the upsampling bandwagon

Reply #18
From the Pro-Audio mailinglist krabapple was referring to I understand that the non-linear-phase apodizing filter modifies the phase response of the output, very likely also in the audible band under 20 kHz. While some people might judge this as an improvement in sound quality, the only way to be sure is to compare (ABX) with a "true-hi-res" version in 24/96 or even 24/192 format. It was suggested on the Pro-Audio list that the phase behaviour of the filter might have a euphonic effect, but in that case it should be called an effect and not a correction of a presumed defective ADC.


Thanks, Kees.  My original version of the opening post here made reference to trade-offs from masking pre-ringing -- I was thinking particularly of the argument from PA that you cite.  I edited it out but was thinking of it again when I criticized Dolby's assumption that their (Meridian's) filter doesn't color the sound, but rather 'restores the natural tonality'.

Quote
What surprises me is that the upsampled content content will be played on 96 kHz DACs with pre-ringing linear-phase filters. Apparently Dolby considers pre-ringing at 96 kHz less of a problem.



It surprised me (and some at PA, including one well-known name who otherwise came to Dolby's defense) that Dolby considers pre-ringing at 48kHz a problem.

Dolby jumps on the upsampling bandwagon

Reply #19
Obviously, elimination of audible preringing could have some benefits. The audible benefits would follow the degree to which this alleged problem exists, if it exists at all.

Not end-user players and receivers, this is the "essence" of the debate (obnoxious font size not quoted).  Let there be no mistaking, like the rest of you I am highly skeptical that this technology will actually provide any tangible benefit (that is also sonically transparent!) for the very reason that Wombat gave.  The problem is that we don't really get very far when we demonstrate woeful reading comprehension with such belligerence.

...and yes, the word "proven" in the excerpt I provided makes me cringe.



I'm afraid the only belligerence I've sensed so far, is yours, old chap, including putting words in my fingers that I never typed.  I was going more for a tone of regretful chiding towards these Dolby claims.  I'll leave it to the gentle reader to decide if I achieved that, and if my reading comprehension is 'woeful'...and to decide who has actually read what.

Dolby jumps on the upsampling bandwagon

Reply #20
I'll leave it to the gentle reader to decide if I achieved that, and if my reading comprehension is 'woeful'...and to decide who has actually read what.

I am *perfectly* comfortable with that.

Dolby jumps on the upsampling bandwagon

Reply #21
From the Pro-Audio mailinglist krabapple was referring to I understand that the non-linear-phase apodizing filter modifies the phase response of the output, very likely also in the audible band under 20 kHz. While some people might judge this as an improvement in sound quality, the only way to be sure is to compare (ABX) with a "true-hi-res" version in 24/96 or even 24/192 format. It was suggested on the Pro-Audio list that the phase behaviour of the filter might have a euphonic effect, but in that case it should be called an effect and not a correction of a presumed defective ADC.
I think it's linear phase up to about 18kHz - if true, I can't imagine many people would hear the phase distortion.

It tries to fall off so that it kills everything by just below Nyquist (including any ringing at/around Nyquist in the sampled signal) while avoiding pre-ringing. I can't imagine why anyone would hear any benefits to this either. I could come up with explanations based around distortion products in real world equipment, but I don't know if it would ever hold true. If it did, you could ABX it. I have my doubts.

Quote
What surprises me is that the upsampled content content will be played on 96 kHz DACs with pre-ringing linear-phase filters. Apparently Dolby considers pre-ringing at 96 kHz less of a problem.
There's no content left there to ring. You can't make a sinc-like filter ring unless you hit it with some content that's close-ish to the transition region. In these signals, they'll be nothing at all, so no ringing.

Cheers,
David.


Dolby jumps on the upsampling bandwagon

Reply #23
I think it's linear phase up to about 18kHz - if true, I can't imagine many people would hear the phase distortion.


I studied the Meridian filter a bit shortly after its appearance. (Hint: a magazine-associated website published its impulse response in numerical form...).

Its frequency response starts rolling off above 18kHz, being -0.5dB at 19kHz.

Phase deviates much earlier, even below 10kHz IIRC. Still, non-linear phase distortion is kept relatively mild.

The standing questions still are:

-is filter pre-ringing at above 20kHz audible?
-what is the detection limit for NL phase distortion in the 10-20kHz octave?



Dolby jumps on the upsampling bandwagon

Reply #24
I think it's linear phase up to about 18kHz - if true, I can't imagine many people would hear the phase distortion.


I studied the Meridian filter a bit shortly after its appearance. (Hint: a magazine-associated website published its impulse response in numerical form...).

Its frequency response starts rolling off above 18kHz, being -0.5dB at 19kHz.

Phase deviates much earlier, even below 10kHz IIRC. Still, non-linear phase distortion is kept relatively mild.

The standing questions still are:

-is filter pre-ringing at above 20kHz audible?


I am aware of no DBTs showing that this is an audible problem in reasonably good commercial equipment, by which I mean modern mainstream Blu ray players selling for > $80 as new equipment. 

Quote
what is the detection limit for NL phase distortion in the 10-20kHz octave?


As long as the phase changes are equally applied to both channels, IME the ear's tolerance for phase distortion is *huge* - over a thousand degrees > 2 Khz. I understand that the ear lacks the ability to meaningfully detect over all phase shift above about 1 KHz.

And that 0.5 db drop at 19 KHz with negligible loss up to say 15 Khz?  DBT all you want, but don't say I didn't warn you!

Brick wall filters @ 16 Khz are generally sonically transparent or very close to it. What can you do to 18 Khz that is worse than that?  ;-)

Yes, there may be some music somewhere where the 16 Khz brick wall is detectable, and I even know about one kind - but it involved a rare folk instrument that produced high level lightly damped sine waves around 16 Khz when it was played. Back in the real world of western music being played the usual ways that the other 99.9999% of the world listens to...