The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
Xiph answers to PONO marketing Digital Converters
Old 6th August 2014
  #91
Motown legend
 
Bob Olhsson's Avatar
 

Verified Member
The fact is that 96k IS more fidelity unless it has been up-sampled from something lower. The only question is if one can easily hear degradation from down-sampling which obviously will vary in quality with different filters, different program material and different listeners because hearing damage reduces each individual's masking ability at certain frequencies which will increase their sensitivity to artifacts in that range.

All you can legitimately say is that it should be able to be done transparently for most people using double precision math and proper dithering. You can't really generalize beyond that because it's a gross over-simplification of a series of engineering processes and of human perception.
Old 6th August 2014
  #92
Lives for gear
 
Plush's Avatar
Quote:
Originally Posted by Jon Hodgson View Post
Interesting you claim his work as your bible (it's an excellent book), and then go on to contradict his stance.

John Watkinson is most definitely not a believer in the need more more than 44.1/16 in a delivery format.

Apple, Beats and fools with money who trust celeb endorsements • The Register
Thank you for the citation. I am sorry to see John Watkinson joining those who complain about Pono.

When I spoke about the wonderful Watkinson tome, The Art of Digital Audio, earlier. I made reference to how it is a complete and detailed primer, analysis and prediction book.
I did not cite it to support my views, only as a learning tool for those serious about understanding digital audio.

Watkinson has written with a different endorsement and conclusion about hi-res in his column in "Resolution" magazine.

I never spoke about or named Watkinson as one who is in my camp of advocating hi-res delivery to rank and file listeners. Notice that I named the names of those who are in my camp with specificity.

I differ with Watkinson and side with Neve in the belief that a 20Hz. to 20kHz. bandwidth is insufficient--most especially when making the recording.
Old 6th August 2014
  #93
Lives for gear
 
Hermetech Mastering's Avatar
 

Verified Member
Anyone who has read both Watkisnon and Pohlmann care to comment on which is the most comprehensive?
Old 6th August 2014
  #94
Lives for gear
 
dcollins's Avatar
 

Verified Member
Watkinson by far. But also more technical.

DC
Old 6th August 2014
  #95
Lives for gear
 
Yannick's Avatar
 

Quote:
Originally Posted by Plush View Post
I differ with Watkinson and side with Neve in the belief that a 20Hz. to 20kHz. bandwidth is insufficient--most especially when making the recording.
Of course this 20-20K bandwidth is of the utmost importance - in analogue component design.
There are NO analogue circuits that are severely bandlimited and transparant.

Digital audio behaves in a completely different way.
a 44.1KHz digital system can reproduce the 0-20K bandwidth PERFECTLY.
The analogue stages of course need (vastly) more bandwidth to be linear in the very same region.
Old 6th August 2014
  #96
Motown legend
 
Bob Olhsson's Avatar
 

Verified Member
Quote:
Originally Posted by dcollins View Post
Watkinson by far. But also more technical.

DC
Old 6th August 2014
  #97
Lives for gear
 

So is there is a practical sampling rate limit, where is it, and why? Is it 192k, 384k, 768k, or beyond, where the high sample rate and marketing people will be satisfied and at peace.

Or should we fix the design problems with filters for 44/48.
Old 6th August 2014
  #98
Motown legend
 
Bob Olhsson's Avatar
 

Verified Member
Certainly design problems ought to be fixed. A number of people believe the ideal for transparent response below 20 k to be around 60 kHz which would be over-sampled at 120k. People I respect have heard better results at 192 with specific converters but that's a matter of overall design that may have little to do with just sample rate. This is one of those "it really depends" topics.
Old 7th August 2014
  #99
Lives for gear
 
Hermetech Mastering's Avatar
 

Verified Member
Quote:
Originally Posted by dcollins View Post
Watkinson by far. But also more technical.

DC
Cheers Dave, look like another book I'll need to add to my shelf!
Old 7th August 2014
  #100
Lives for gear
 

Quote:
Originally Posted by Bob Olhsson View Post
The fact is that 96k IS more fidelity ...
True, only for signals that contain significant energy above 20kHz?
Old 7th August 2014
  #101
Lives for gear
 

Verified Member
Quote:
Originally Posted by David Spearritt View Post
True, only for signals that contain significant energy above 20kHz?
It's true for signals that contain anything above where the effects of the anti aliasing filter become significant, whatever that happens to be.

However, being higher fidelity to the total original signal doesn't necessarily mean any higher fidelity to that information in the signal that we can actually perceive.
Old 7th August 2014
  #102
Moderator
 
narcoman's Avatar
 

Quote:
Originally Posted by Plush View Post
Thank you for the citation. I am sorry to see John Watkinson joining those who complain about Pono.
My main complaint is that its a money rinse. Or an attempted one. Streaming services are moving to 24bit and higher sample rates anyway - we don't need Pono.


Quote:
Originally Posted by Plush View Post
I differ with Watkinson and side with Neve in the belief that a 20Hz. to 20kHz. bandwidth is insufficient--most especially when making the recording.
I agree
Old 7th August 2014
  #103
Deleted User
Guest
Quote:
Originally Posted by David Spearritt View Post
So is there is a practical sampling rate limit, where is it, and why?
Quote:
Originally Posted by Bob Olhsson View Post
This is one of those "it really depends" topics.
One example, capturing analogue tape bias which can range anywhere from 30kHz up through to 450kHz. Since tape bias varies dramatically and can not always be determined a priori, which sample rate will be required to accurately capture this metadata? None of the proposed rates suffice.

The CD-4 format has a carrier of around 50 kHz and again, in order to capture and preserve this metadata, a higher sampling rate is required. There is no skirting the issue.

Just two examples to consider, which relate why DSD256 and DSD512 are, in my opinion, a step in the right direction from DSD128 and in no way because they sound any better or because using those rates will move more units sold.

I don't follow the 60kHz "optimal sample" rate pushers. Optimal for what? Certainly that rate is not optimal for the use cases cited above. It's not always about what we can hear, often it's about what we can't and what we lose by disregarding something clearly inaudible but which often proves rather significant.

cheers,
Reynaud
Old 7th August 2014
  #104
Lives for gear
 

Verified Member
Quote:
Originally Posted by Deleted User View Post
One example, capturing analogue tape bias which can range anywhere from 30kHz up through to 450kHz. Since tape bias varies dramatically and can not always be determined a priori, which sample rate will be required to accurately capture this metadata? None of the proposed rates suffice.

The CD-4 format has a carrier of around 50 kHz and again, in order to capture and preserve this metadata, a higher sampling rate is required. There is no skirting the issue.

Just two examples to consider, which relate why DSD256 and DSD512 are, in my opinion, a step in the right direction from DSD128 and in no way because they sound any better or because using those rates will move more units sold.

I don't follow the 60kHz "optimal sample" rate pushers. Optimal for what? Certainly that rate is not optimal for the use cases cited above. It's not always about what we can hear, often it's about what we can't and what we lose by disregarding something clearly inaudible but which often proves rather significant.

cheers,
Reynaud
Why would you want to capture tape bias? It's there to linearize the performance of the tape in the audio band, the whole point is that in itself it's not audible, it's not even metadata since it provides no useful information, it's just a leftover.

Also bear in mind that most of that DSD bandwidth is so full of noise (shaped up from the lower bands) no original signal will survive.
Old 7th August 2014
  #105
Deleted User
Guest
Quote:
Originally Posted by Jon Hodgson View Post
Why would you want to capture tape bias? It's there to linearize the performance of the tape in the audio band, the whole point is that in itself it's not audible, it's not even metadata since it provides no useful information, it's just a leftover.
Firstly, because it is crucial metadata that needs to be retained and assigned to assets. Secondly because it is useful in correcting drift, wow and flutter. Currently this is performed in the analogue domain due to the limits on commercially available systems to capture these signals.

Quote:
Originally Posted by Jon Hodgson View Post
Also bear in mind that most of that DSD bandwidth is so full of noise (shaped up from the lower bands) no original signal will survive.
Which is why I specifically mentioned DSD256 and DSD512 as being an improvement.

cheers,
Reynaud
Old 7th August 2014
  #106
Lives for gear
 

Verified Member
Quote:
Originally Posted by Deleted User View Post
Firstly, because it is crucial metadata that needs to be retained and assigned to assets. Secondly because it is useful in correcting drift, wow and flutter. Currently this is performed in the analogue domain due to the limits on commercially available systems to capture these signals.
I don't see how it is "crucial metadata that needs to be retained and assigned to assets". For most of the people on this forum, it's of no use at all, if they're using tape it's because they want that sound, wow and flutter included.

I can see that in converting old recordings to digital it can be useful as part of the process, in order to eliminate wow and flutter, but once that is done does it have any use?

So for the specific task of converting archive material I can see that you'd want a conversion system that captured a far greater bandwidth than audio with enough SNR to extract a bias signal from it, but single bit DSD is a poor choice, no matter what the oversampling ratio, better to use a mulibit oversampled noise shaped quantizer (that will give you a far better signal to noise ratio), extract the bias signal from the raw signal, then downsample it to an ordinary PCM file, with the bias information as actual metadata (or use that metadata to perform the wow and flutter compensation at the same time as the downsampling).
Old 7th August 2014
  #107
Moderator
 
narcoman's Avatar
 

Quote:
Originally Posted by Deleted User View Post
I don't follow the 60kHz "optimal sample" rate pushers. Optimal for what? Certainly that rate is not optimal for the use cases cited above. It's not always about what we can hear, often it's about what we can't and what we lose by disregarding something clearly inaudible but which often proves rather significant.

cheers,
Reynaud
But thats more usually technical data.

The 60kHz rate is bounded about as being economic, well within technical means (even more so at the time of the writing of the white paper you're referring to) and sufficiently far enough away from hearing bounds to mean that all audible input is well away from brick wall filters.

This has nothing to do with capturing OTHER data such as bias information. That is only relevant in a small amount of cases; capturing tape recorded performance where it MAY have some use in the future.

Now, citing the oft held premise that there is all kind of information "up there" which we use as humans to hear may or may not hold some argument water, although not for me.

I still think that Jazz Samba from Stan Getz et al is one of the finest recordings ever made.... and its pretty far from optimal or real sounding or "right". Sufficiency is always a test and one or more of the current PCM formats IS sufficient for the broad enjoyment of even high end recording. All other formats are, for me, either great alternatives or just re-inventing the wheel. There just isn't much point in it for music apart from for those who which to push an agenda way beyond sufficiency (which has merits but will not be a mainstay on human consumption).
Old 7th August 2014
  #108
Deleted User
Guest
Quote:
Originally Posted by narcoman View Post
The 60kHz rate is bounded about as being economic, well within technical means (even more so at the time of the writing of the white paper you're referring to)
I was not referencing a particular paper or author as 3M and Soundstream were pushing a similar rate (50kHz if I recall) in the early to mid-70s. Any subsequent paper is merely agreeing with their conclusions and subsequent stance, right or wrong.

Quote:
Originally Posted by narcoman View Post
This has nothing to do with capturing OTHER data such as bias information. That is only relevant in a small amount of cases; capturing tape recorded performance where it MAY have some use in the future.

Now, citing the oft held premise that there is all kind of information "up there" which we use as humans to hear may or may not hold some argument water, although not for me.
If you will be careful to notice, I was responding to a (possibly) rhetorical question presented by David Spearritt. I specifically avoided anything to do with "quality", "sufficiency" (this comes in to play below), "perfection" or the like because that was not, in my view, part of the question as originally posed.

Yes, I concede that the majority of engineers and innocent bystanders on this forum may not care or have any interest in bias signals, or ultrasonic and infrasonic content but that does not diminish their importance. The majority is frequently not the best judge of anything important.

Quote:
Originally Posted by narcoman View Post
Sufficiency is always a test
Really? For how long? New data, new tests, and new requirements will change the parameters of sufficiency. Again, sufficient for whom and in which context?

Quote:
Originally Posted by narcoman View Post
one or more of the current PCM formats IS sufficient for the broad enjoyment of even high end recording
Apparently so is mp3. Very sufficient if one does any meaningful data capture of "news server" usage - where, incidentally, a particular "classical" DSD-only title seemed rather popular in mp3 format (the wav version was barely touched and the "DSD" version had less than a handful of downloads). Sufficient indeed.

Quote:
Originally Posted by narcoman View Post
All other formats are, for me, either great alternatives or just re-inventing the wheel. There just isn't much point in it for music apart from for those who which to push an agenda way beyond sufficiency (which has merits but will not be a mainstay on human consumption).
Again, sufficient for whom. mp3 seems more than sufficient for a significant portion of the population. Which definition of sufficiency is advisable, yours or theirs?

Decca settled on 18/48k in the 70s and I don't think the reason those recordings are so wonderful is because of the chosen rate. Perhaps good and well rounded engineering with a good wallop of experience may be the major contribution. People forget that Decca was all digital in the late 70s (some even as early as 1971) and that every Decca CD release has a lower rate than the original capture. Did people throw up their arms in disillusionment? 16bit/44.1kHz seems sufficient indeed, until you consider those mp3 listening fiends of quality.

cheers,
Reynaud
Old 7th August 2014
  #109
Moderator
 
narcoman's Avatar
 

Its market driven. So sufficient for market.

MP3, as horrible as they are (and I don't even like CD) , are sufficient for market needs. We can wax lyrical about artistry all we like - no matter which way you cut it market forces define sufficiency. MP3 was never something I wanted, and certainly not the 128 level that the market initially settled on. I also didn't want hyper limited mixing/masters. But thats what the market was drawn towards... rightly or wrongly. This very much files under the zone of "never underestimate the general markets desires"!!!

The highest possible technical format over the most economic realistic delivery mechanism that does leave 90% of current hardware redundant is where a market will always temporally float. If there is a market to sell 24 bit 192 masters (which I very very much doubt) then it won't need Pono (i know you aren't talking about that right now but it's my closer )
Old 7th August 2014
  #110
Lives for gear
 

Verified Member
Quote:
Originally Posted by Deleted User View Post
I was not referencing a particular paper or author as 3M and Soundstream were pushing a similar rate (50kHz if I recall) in the early to mid-70s. Any subsequent paper is merely agreeing with their conclusions and subsequent stance, right or wrong.
Not exactly, the factors used to calculate the sample rate can vary.

For example Dan Lavry bases his 60kHz figure on a 20kHz audible bandwidth and what he considers to be insignificant passband ripple.

On the other hand I've seen papers (I think from Meridian) which use a somewhat higher bandwidth for what is audible (I think something like 24kHz) because in tests a few individuals have demonstrated perception over 20kHz, so I presume they use different figures when it comes to the effects of filtering.

Quote:
Yes, I concede that the majority of engineers and innocent bystanders on this forum may not care or have any interest in bias signals, or ultrasonic and infrasonic content but that does not diminish their importance.
No, but it does illustrate their importance, which in most cases is zero.
IF you want to be able to counter wow and flutter, THEN capturing a greater bandwidth that includes bias information is useful, otherwise it's pointless.
Old 7th August 2014
  #111
Deleted User
Guest
Quote:
Originally Posted by narcoman View Post
If there is a market to sell 24 bit 192 masters (which I very very much doubt) then it won't need Pono (i know you aren't talking about that right now but it's my closer )
Well in certain markets even Pono falls short.

nativedsd is pushing all sorts of hardware that already ensures Pono is irrelevant in their market even before release. This market appears more than willing to pay similar pricing for the convenience of downloads to the physical release.

Qobuz is apparently moving to launch 24bit/192kHz downloads next month (I have heard the amusing "Studio Master" term thrown about). They already stream in 16bit/44.1kHz FLAC which is growing in popularity.

The market is apparently there and neither service references Pono (even once).

cheers,
Reynaud
Old 7th August 2014
  #112
Lives for gear
 
DSD_Mastering's Avatar
 

Verified Member
Quote:
Originally Posted by Jon Hodgson View Post
Why would you want to capture tape bias? It's there to linearize the performance of the tape in the audio band, the whole point is that in itself it's not audible, it's not even metadata since it provides no useful information, it's just a leftover.
.
For this:

Plangent Process
Old 7th August 2014
  #113
Quote:
Originally Posted by DSD_Mastering View Post
Very interesting, thanks for the link!
Old 7th August 2014
  #114
Motown legend
 
Bob Olhsson's Avatar
 

Verified Member
The 60 kHz. rate is for transparency below 20kHz. Higher has real world parts precision issues according to Dan Lavry and several others.

I've heard DSD beat 192k converters and I've heard other 96k converters handily beat a Korg DSD. Implimentation seems to matter way more than format when it comes to this stuff.
Old 7th August 2014
  #115
Deleted User
Guest
Quote:
Originally Posted by Jon Hodgson View Post
I don't see how it is "crucial metadata that needs to be retained and assigned to assets".
I should have been clearer, it is descriptive metadata. All metadata is relevant.

Quote:
For most of the people on this forum, it's of no use at all, if they're using tape it's because they want that sound, wow and flutter included.
Tape is one of the most reliable and proven archival mediums we currently have access to. Digital storage mechanisms, especially plastics, are not reliable long term storage mediums as they require constant migration. Often, and contrary to your assumption, the "sound" of analogue tape has nothing to do with its use.

Again, I wouldn't use "most of the people on this forum" as a barometer.

Quote:
I can see that in converting old recordings to digital it can be useful as part of the process, in order to eliminate wow and flutter, but once that is done does it have any use?
Further along, an improved system will present itself and retaining that descriptive metadata may prove critical. I capture as much descriptive metadata as possible, even if it may seem irrelevant today.

Migration to new tape stock can also be achieved completely in the analogue domain, correcting wow and flutter in the process and occurs more frequently than you may realise. The source, even in this case is not completely discarded even if a migration is successful and to the highest quality and standards we currently are able to achieve.

The Plangent Process linked to above is but one commercially available system. I am aware of at least two systems that are currently moving through the European patent system, neither of which are currently intended for commercial use. Both are to be employed by the institution's system developers and supporting staff.

No current system is perfect in all respects. Anyone to claim such is deluding themselves. If there were, there surely would be no need for alternative systems.

Quote:
Originally Posted by Jon Hodgson View Post
No, but it does illustrate their importance, which in most cases is zero.
IF you want to be able to counter wow and flutter, THEN capturing a greater bandwidth that includes bias information is useful, otherwise it's pointless.
My experience says otherwise. Often by moving beyond what is expected, by pushing to do more than what is required it has only better served the client and their expected result further down the line within a new set of parameters and requirements. Where my gut wins the day and not by using others as the barometer of how best to serve someone.

cheers,
Reynaud
Old 7th August 2014
  #116
Moderator
 
narcoman's Avatar
 

Quote:
Originally Posted by Deleted User View Post
Well in certain markets even Pono falls short.

nativedsd is pushing all sorts of hardware that already ensures Pono is irrelevant in their market even before release. This market appears more than willing to pay similar pricing for the convenience of downloads to the physical release.

Qobuz is apparently moving to launch 24bit/192kHz downloads next month (I have heard the amusing "Studio Master" term thrown about). They already stream in 16bit/44.1kHz FLAC which is growing in popularity.

The market is apparently there and neither service references Pono (even once).

cheers,
Reynaud
I agree there are products - but there is scant interest in public buy in. Well - thats my view anyway!!
Old 7th August 2014
  #117
Lives for gear
 

Verified Member
Quote:
Originally Posted by Deleted User View Post
I should have been clearer, it is descriptive metadata. All metadata is relevant.
Not if you're not going to use it.
Quote:
Tape is one of the most reliable and proven archival mediums we currently have access to. Digital storage mechanisms, especially plastics, are not reliable long term storage mediums as they require constant migration. Often, and contrary to your assumption, the "sound" of analogue tape has nothing to do with its use.
It's not contrary to my assumption, "for most of the people on this forum" means exactly that, not all people everywhere. Discussions about tape on this forum are generally about sound (plus there are many about tape sound emulations), and occasionally about workflow, medium longevity is mentioned on occasion, but very rarely in comparison.

As for the reliability of digital storage solutions, these days options are available which have the potential to far exceed that of analogue tape, because the world runs on data and so the storage and preservation of it has become vital, as a result anybody in the world can now upload an audio file to a server and have it replicated in multiple locations worldwide, backed up and shifted to new media as it gets installed. The data bits on the individual medium (e.g. a magnetic drive) may not be as resistant to deterioration as an analogue signal on tape, but the complete storage solution can be far more so.
Quote:
Again, I wouldn't use "most of the people on this forum" as a barometer.
Nor would I use the niche areas you seem to be.
Quote:
Further along, an improved system will present itself and retaining that descriptive metadata may prove critical. I capture as much descriptive metadata as possible, even if it may seem irrelevant today.
If your intention is to recover the original signal pre-recording to the greatest degree possible, then this makes perfect sense. Once again though, this does not seem to be an aim or requirement of the vast majority of people capturing digital audio, a large proportion of which has never been near tape, and of that which has, a great deal is using tape in effect as a processing medium.
Quote:
Migration to new tape stock can also be achieved completely in the analogue domain, correcting wow and flutter in the process and occurs more frequently than you may realise. The source, even in this case is not completely discarded even if a migration is successful and to the highest quality and standards we currently are able to achieve.

The Plangent Process linked to above is but one commercially available system. I am aware of at least two systems that are currently moving through the European patent system, neither of which are currently intended for commercial use. Both are to be employed by the institution's system developers and supporting staff.

No current system is perfect in all respects. Anyone to claim such is deluding themselves. If there were, there surely would be no need for alternative systems.
I will concede, that in the specific situation of dealing with archive material originally recorded to tape, keeping the bias information can have potential benefits, but that's really just a fraction of the use of digital audio, and we're talking archiving not distribution. Pono and other similar systems are about distribution.

What you're talking about is fascinating stuff, but niche, and in my opinion best dealt with by appropriate capture hardware.
Quote:
My experience says otherwise. Often by moving beyond what is expected, by pushing to do more than what is required it has only better served the client and their expected result further down the line within a new set of parameters and requirements. Where my gut wins the day and not by using others as the barometer of how best to serve someone.
So apart from the specific use of bias signal in correcting audio tape transfers, do you have a concrete example of capturing ultrasonics actually being of benefit down the line?
Old 8th August 2014
  #118
Lives for gear
 
Jerry Tubb's Avatar
 

Verified Member
24-bit is the big leap in quality IME.

Best, JT
Old 8th August 2014
  #119
Lives for gear
 
AlexK's Avatar
 

RE the higher sample rates thing - I run at 88.2 kHz simply because I've found most common converters (192s, Lynx for example, Apogee AD16 etc) seem to sound noticeably better at those sample rates. I don't think it's because of the sample rate per se, it just gets the aliasing & filtering issues way out of the passband which to be honest, are still far from perfect on many of these older converters. Modern AD doesn't seem to sound so different at different sample rates. An Orion at 44.1k seems to sound the same as it does at 88.2 or 96.

One advantage of higher sample rates is being able to do Varispeed, sometimes quite extreme variations without encountering much lost top-end. Because I'm generally doing more than just engineering most of the projects I work on and I end up getting quite creative, having that extra HF information has occasionally come in useful.
Old 8th August 2014
  #120
Quote:
Originally Posted by mixmixmix View Post
On a large scale 24/192 downloads won't survive as a premium priced service because consumers will realize at some point that there is no audible differences between 24/192 and 16.44.

However small croud of hi-rez enthusiasts may be able to support tiny hi-rez industry.

Whilst I personally consider 24/192 overkill for a variety of reasons, the statement above is typically born of complete & utter ignorance, cloth ears or lousy monitoring. The difference between 16/44.1 and 24/96 is almost night & day to me.
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump