Taperssection.com

Gear / Technical Help => Playback Forum => Topic started by: MakersMarc on March 29, 2020, 09:31:03 PM

Title: 24/192
Post by: MakersMarc on March 29, 2020, 09:31:03 PM
Who can hear 24/192? Can any human. And my DAC only does 24/96, am I even missing anything? Trying to decide whether to kick down for 24/192 downloads of the GD ‘76 box. I want to support the Band but no more fucking box sets, no room.😂
Title: Re: 24/192
Post by: jerryfreak on March 29, 2020, 11:01:26 PM
ill pay you $100 if you can demonstrate via blind abx that you can hear a difference between 24/48 and 24/192
Title: Re: 24/192
Post by: noahbickart on March 30, 2020, 08:53:08 AM
I makes me feel better about my audiophile street cred when I see lots of fancy lights go on on my dac.

When using dynamics and eq plugins there might be a slight advantage to higher sample rates.

But no, listening normally even on a fancy system, I can’t hear the difference
Title: Re: 24/192
Post by: KenH on March 31, 2020, 10:43:44 AM
I think the 24/192 GD releases sound *amazing*...  so smooth and detailed.
Title: Re: 24/192
Post by: Gutbucket on March 31, 2020, 11:49:48 AM
^ Which has a strong correlation to the work put into them, and a weak correlation to the sampling rate of the release
Title: Re: 24/192
Post by: KenH on March 31, 2020, 12:01:14 PM
^ in your opinion...  I hear what I hear, and I don't see how the work of mastering (besides offering the higher sample rate) is any different, from, say 2mastering 24/96...  it's just a different setting on the AD conversion unit, which produces a better result, imho.
Title: Re: 24/192
Post by: Gutbucket on March 31, 2020, 12:20:21 PM
maybe
Title: Re: 24/192
Post by: Gutbucket on March 31, 2020, 12:32:12 PM
Not disputing whether they sound amazingly smooth and detailed..
but questioning attribution of those aspects without clear evidence and multiple variables in play.

We do not know the details of the mastering.
As Noah mentions, it's not necessarily just a different output setting.
Title: Re: 24/192
Post by: EmRR on March 31, 2020, 01:00:21 PM
ill pay you $100 if you can demonstrate via blind abx that you can hear a difference between 24/48 and 24/192

The only place I've ever been able to hear it has been in a John Arthur designed pro mastering room on a set of Dunlavey SC-V's driven by Pass Labs amps.  A room that convincingly acts like a big set of headphones when you're in the sweet spot.   I'd win that bet in that room.  Elsewhere, very doubtful. 

There are lot of layers to parsing the value of it, most of it is on the mixing and processing side, delivery mot as much.  But then you get into what the converter sounds like, what the system is like, the room, etc.  Lots of C-hair level nuances that may all add up to something.....or not.....
Title: Re: 24/192
Post by: noahbickart on March 31, 2020, 01:25:21 PM
What’s the frequency response of the original tapes? There’s no way there’s any musical info above 20khz. In which case, 44.1khz is all you’d need. I don't know about you, buy I'm 43. I can't hear over 20k and probably never could.

I’d wager that there’s no more than 80db of dynamic range there either. In which case the 16 bit version is plenty good.

I get why it makes sense to record (even from the master tapes) at 24/192. I'd do that too. But for playback?

I download the 24/192 and then do my own 192>48 conversion. That's good enough for me.

I have a problem with the Phish releases on this level as well. Why should content which came from a 16/48khz DAT tape ever be sold at 24bit?
Title: Re: 24/192
Post by: KenH on March 31, 2020, 02:13:49 PM
Have any of you actually heard any of the 24/192 GD releases ?  Granted, I haven't done an A/B comparison, but they're the best I've ever heard...

IMHO, it's the sample rate that I think makes all the difference.   44.1/48 kHz vs 192 kHz _should_ make a difference, and be closer to the analog masters themselves.
Title: Re: 24/192
Post by: Gutbucket on March 31, 2020, 03:00:02 PM
In what ways should it be closer to the analog source?

What leads you to believe playback sample rate makes all the difference?

Title: Re: 24/192
Post by: Gutbucket on March 31, 2020, 03:03:00 PM
The only place I've ever been able to hear it has been in a John Arthur designed pro mastering room..

I'd love to experience that room!
Title: Re: 24/192
Post by: KenH on March 31, 2020, 03:05:30 PM
Analog waves are a continuous, smooth curve.  Digital samples are steps, like a basic calculus concept.   The more samples, the more you approximate the continuous smooth curve of the analog sound wave.   The extreme analogy on the opposite spectum is MP3 sound with less samples per second (and even then each sample is compromised with omissions in each sample).
Title: Re: 24/192
Post by: aaronji on March 31, 2020, 03:19:15 PM
Except that's not how it works for a bandwidth limited source sampled at the Nyquist frequency...
Title: Re: 24/192
Post by: Gutbucket on March 31, 2020, 03:19:51 PM
The output of a DAC is analog with no steps.
Title: Re: 24/192
Post by: EmRR on March 31, 2020, 03:20:17 PM
What’s the frequency response of the original tapes? There’s no way there’s any musical info above 20khz. In which case, 44.1khz is all you’d need. I don't know about you, buy I'm 43. I can't hear over 20k and probably never could.

I have a problem with the Phish releases on this level as well. Why should content which came from a 16/48khz DAT tape ever be sold at 24bit?

It's not about frequency response, it's much more about phase response and intermodulation distortion. 

Phish DAT's : as soon as you add any processing at all, you've created a 24 bit word. 

These new Dead releases:  anyone know if they applied Plangent processing to hi-res files?  That'd make a HUGE difference in the way they sound, regardless of delivery version. 

I think if they did the work at hi-res, there will always be people who want to have it that way, so it's nice it's made available.

Title: Re: 24/192
Post by: dyneq on March 31, 2020, 03:21:22 PM
I have ABX'd a fair amount and cannot hear a difference between 16/44 and 24/48 (I'm in my mid-50s, so take that in to account), so my opinion is that 24/192 is a solution in search of a problem.

I highly recommend that folks do ABX testing for themselves. I use foobar2000's ABX Comparator, but there is also a web tool here (http://lacinato.com/cm/software/othersoft/webabx). I have also used ABX to determine the best audio compression level to use in noisy, on-the-go environments (exercising, car, etc).

Luckily for me, dead.net sells ALACs at 16/44 for less $ which I convert to FLAC.
Title: Re: 24/192
Post by: Gutbucket on March 31, 2020, 03:40:57 PM
Analog waves are a continuous, smooth curve.  Digital samples are steps, like a basic calculus concept.   The more samples, the more you approximate the continuous smooth curve of the analog sound wave.   The extreme analogy on the opposite spectum is MP3 sound with less samples per second (and even then each sample is compromised with omissions in each sample).

Lossy compression schemes such as mp3 are not relevant to this discussion.

"like a basic calculus concept" - Ok, Serious mathematical question: Is the derivation of the area under the curve using calculus the actual area under the curve or not?  Perhaps you believe in calculus but not in sampling theorem?

"The more samples, the more you approximate the continuous smooth curve of the analog sound wave."

This is true, yet in a way that is different than you think.

With respect to playback (which is important to clarify), a digital representation fully recreates the smooth continuous curve of the original analog sound wave up to about 1/2 the sampling rate.  A higher sample rate extends that to higher frequencies.  It does not make the representation up to 1/2 the lower sampling rate "more accurate".

^This is the essence of sampling theory.

The phase-response and intermodulation points EmRR brings up come are why working at higher sampling rates is relevant in the mixing/processing/mastering stages.

..anyone know if they applied Plangent processing to hi-res files?  That'd make a HUGE difference in the way they sound, regardless of delivery version.

Agreed.  This would make a significant audible difference.
Title: Re: 24/192
Post by: jefflester on March 31, 2020, 04:06:14 PM
These new Dead releases:  anyone know if they applied Plangent processing to hi-res files?  That'd make a HUGE difference in the way they sound, regardless of delivery version. 
Yes, but the Plangent Process is part of the A/D step. So the 16/44.1 files get the same advantage.
Title: Re: 24/192
Post by: morst on March 31, 2020, 04:08:25 PM
These new Dead releases:  anyone know if they applied Plangent processing to hi-res files?  That'd make a HUGE difference in the way they sound, regardless of delivery version. 
I assume they do all their analog releases via Plangent. Plangent happens to require 192 k sampling in order to track the original 56k bias tone, so the masters MUST be ripped from analog at 192.
I'm personally not able to hear much over 10k, so I would definitely not be into recording at rates beyond 48kHz unless I plan to be using the audio in a pitch shifting (sampling) situation. Certainly not for live music.
But I'm just an old human.
Some people might be purchasing the disc set for a pet bird or dog?
Title: Re: 24/192
Post by: KenH on March 31, 2020, 04:10:48 PM
Analog waves are a continuous, smooth curve.  Digital samples are steps, like a basic calculus concept.   The more samples, the more you approximate the continuous smooth curve of the analog sound wave.   The extreme analogy on the opposite spectum is MP3 sound with less samples per second (and even then each sample is compromised with omissions in each sample).
Lossy compression schemes such as mp3 are not relevant to this discussion.

I think it's analogous, no pun intended...  Was just thinking that MP3s are at the lower end of wave approximation vs higher sample rates and bit sample size.

"like a basic calculus concept" - Ok, Serious mathematical question: Is the derivation of the area under the curve using calculus the actual area under the curve or not?  Perhaps you believe in calculus but not in sampling theorem?

The area under the curve is an approximation in calculus unless you reach infinity (though it's been almost 40 years (!) since I studied calculus as a math major).  Even the area of a circle, using Pi, which is not an exact number, is a [close] approximation, no ?

I haven't studied sampling theorem, but do know at least some of the concepts.  The more samples you take, the _closer_ you get to the original wave form.

"The more samples, the more you approximate the continuous smooth curve of the analog sound wave."

This is true, yet in a way that is different than you think.

With respect to playback (which is important to clarify), a digital representation fully recreates the smooth continuous curve of the original analog sound wave up to about 1/2 the sampling rate.  A higher sample rate extends that to higher frequencies.  It does not make the representation up to 1/2 the lower sampling rate "more accurate".

^This is the essence of sampling theory.

OK, I sort of follow that, analog by definition is a continuous wave.   Not sure I follow the latter part of your paragraph.   Where I mostly notice the difference in 24/192 is with the cymbals, for sure.
Title: Re: 24/192
Post by: aaronji on March 31, 2020, 04:46:55 PM
The area under the curve is an approximation in calculus unless you reach infinity (though it's been almost 40 years (!) since I studied calculus as a math major).

The integral is the limit as the width of the theoretical rectangles (Δx) approaches zero, right? So, when the width is infinitesimally small, the calculated area under the curve is exact (to the extent that the curve can be defined by a function).

Read this white paper from Dan Lavry (http://lavryengineering.com/pdfs/lavry-sampling-theory.pdf). It has been posted here many times, but it does a good job of explaining why sampling theory is exact for bandwidth limited sources (like those made on tape) sampled at/above the Nyquist frequency.
Title: Re: 24/192
Post by: Gutbucket on March 31, 2020, 05:31:25 PM
I see the analogy you are making with mp3.  The difference is perceptual codecs are designed to intentionally discard information within the audible range deemed to be perceptually unimportant, whereas a straight PCM representation attempts exact reproduction within its bandwidth limitations.  Higher sampling rates and greater bit depths further extends that bandwidth.  Extending bandwidth may be directly audible in some cases as discussed above, but will generally be more useful for carrying out the calculations performed by mixing, EQ, dynamics and other processes to a greater degree of precision, which is far more likely to be audible.

Nyquist Sampling Theorem states that a bandlimited (meaning: choose a high-frequency limit) continuous-time signal (smooth analog waveform) can be sampled (measured/recorded) and then perfectly reconstructed from that sample data if the waveform is sampled over twice as fast as it's highest frequency component (reconstruction up to 1/2 the sampling rate - somewhat less in actualized systems to make allowance for real-world bandwidth-limitation filters).

Calculation of the area of a circle using Pi is exact.. the problem becomes the approximation of Pi, how many decimals do we deem to be enough?  Digital sampling works perfectly in a mathematical sense.

All real-world engineering solutions are imperfect. All analog systems are real-world engineering solutions that have bandwidth limitations and measurable inaccuracies within those bandwidth limitations.  How much perfection is needed in any imperfect real-world analog system?  How much is needed in an imperfect real-world digital system?  How much excess digital bandwidth is required to perfectly reproduce an imperfect bandwidth-limited analog system, including its imperfections? When do overkill amounts of "enough" become ridiculous.. and then become counter-productive by imposing other real-world limitations?

If a particular DAC does a better real-world job at one sampling rate versus another, does that mean we should attribute the difference to the sampling rates or to the DAC itself? 

It's a very complex chain of causality. I'm not trying to be argumentative, only warning against reaching a conclusion without considering the full complexity of the problem.

Edit- ..and I suspect that it is "something other than the playback sample rate itself" that is responsible for the very real differences you are hearing.
Title: Re: 24/192
Post by: Gutbucket on March 31, 2020, 06:18:04 PM
If a band or label can make more money by releasing the same material at various sample rates, more power to them.
If people derive more enjoyment from those releases for whatever reason, more power to them.

However, a problem with that is we can rarely if ever be certain that the material is actually the same except for its sample rate.  Even if it actually is (at least to the best of the ability of the releaser) our real-world playback systems may not be handling them equally well.  How do we know which of those things are actually responsible for the difference we hear?

How much falls into expectation bias? Not with regards to listening (I'm not questioning what anyone is hearing), but with regards to everything else including equipment design resources, sales, mastering efforts, etc.  Manufacturers as well as music producers have a strong economic incentive to put more time and effort into making their "high resolution" gear and product sound better than than their "standard resolution" gear and output, even if that is out of the same piece of equipment.  And that is with benign intent.

The darker flip-side are the numerous high-resolution releases that demonstrably had exactly the same bandwidth as the original CD releases, re-released and re-sold at additional/higher profit due to the widely held assumption that the "high resolution" version must be better simply because of the release format.  That is evident by examples of the DVDA/SCAD re-release era of the big labels where folks did a inverted-polarity null-out test with the original CD and found no change.  Remastered versions will show a difference, but do not necessarily require more bandwidth as long as it all still fits comfortably.
Title: Re: 24/192
Post by: EmRR on March 31, 2020, 07:38:52 PM
These new Dead releases:  anyone know if they applied Plangent processing to hi-res files?  That'd make a HUGE difference in the way they sound, regardless of delivery version. 
Yes, but the Plangent Process is part of the A/D step. So the 16/44.1 files get the same advantage.

  That's what I said, more or less.  But Plangent is AFTER A/D and requires high sample rate to capture tape bias tone, so 192 or 384. 
Title: Re: 24/192
Post by: EmRR on March 31, 2020, 07:58:08 PM
If a particular DAC does a better real-world job at one sampling rate versus another, does that mean we should attribute the difference to the sampling rates or to the DAC itself? 

It's a very complex chain of causality. I'm not trying to be argumentative, only warning against reaching a conclusion without considering the full complexity of the problem.

Edit- ..and I suspect that it is "something other than the playback sample rate itself" that is responsible for the very real differences you are hearing.

Bingo!  Looking at just sample rate ignores a ton of other aspects, and end user playback-only is totally different from professional processing needs.

My studio experience with the converters I had was that my previous type made cymbals and acoustic instruments sound very indistinct at 44K1/48, with cymbals particularly trashy.  2005 tech.  Sounded fine at higher rates.  My newer converters (2014 tech) don't care so much, they still sound better at 88K2 than 48, but I have to work to hear it.  We're talking about the same sample rates, and different equipment.  There was one long standard in studios that seemed to sound best at 44K1, if I recall the conversations correctly. 

Apple devices, Quicktime is running everything in the background in iTunes and converting various rates on the fly to whatever you have your master clock set for.  It's not changing rates based on native file rate.  True for many home music servers too.  Experiment with listening to a track at it's native rate, and then at others.  See if you hear any difference, and if you prefer one.  If you prefer a high master rate on a file with a low native rate, then it most likely means that higher rate sounds better on that particular equipment.  Or vice versa. 

Take a high rate file and downconvert it.  Compare.   Pick something like pretty clean acoustic music.   If you pick a loud rock band track, you may like the lower rate, as some people feel more sense of presence with that, and it's preferable on something that's supposed to kick you in the stomach anyway. 

And generally, again, it's not about frequency response in adult humans.  Phase, intermodulation, and stereo image cue timing at high frequencies are more what it's about.   Intermod, you probably won't hear at all on a rock band track, but you sure as hell will on a solo grand piano track if there's enough of it.  Phase distortion in the top will change the perception of harmonics on acoustic stringed instruments, something like a string quartet can be most revealing.   An old rock band tape may sound more or less forward in the top end from phase distortion, and be more of an 'eh' difference. 

Title: Re: 24/192
Post by: jefflester on March 31, 2020, 08:41:50 PM
These new Dead releases:  anyone know if they applied Plangent processing to hi-res files?  That'd make a HUGE difference in the way they sound, regardless of delivery version. 
Yes, but the Plangent Process is part of the A/D step. So the 16/44.1 files get the same advantage.

  That's what I said, more or less.  But Plangent is AFTER A/D and requires high sample rate to capture tape bias tone, so 192 or 384.
I just meant no Plangent advantage of 24/192 over the 16/44.1 for a release like the current June '76 box since they go through the same work flow, which is what I thought the discussion was about. But I'll back out since I don't have much more than a general knowledge on the subject.
Title: Re: 24/192
Post by: EmRR on March 31, 2020, 10:13:34 PM
These new Dead releases:  anyone know if they applied Plangent processing to hi-res files?  That'd make a HUGE difference in the way they sound, regardless of delivery version. 
Yes, but the Plangent Process is part of the A/D step. So the 16/44.1 files get the same advantage.

  That's what I said, more or less.  But Plangent is AFTER A/D and requires high sample rate to capture tape bias tone, so 192 or 384.
I just meant no Plangent advantage of 24/192 over the 16/44.1 for a release like the current June '76 box since they go through the same work flow, which is what I thought the discussion was about.

We're saying the same thing.  No worries. Ships in the night and all that.   Hi rate capture, Plangent process, then you can downsample to whatever because that part's done.  Or you can keep it hi-res for the audiophiles who want all the data. 
Title: Re: 24/192
Post by: MakersMarc on April 15, 2020, 05:50:35 PM
Thanks all, ended up getting the box. I don’t even have a DAC that goes beyond 24/96, and oy, the storage needs. And I’m a sucker for the goodies in the box.

Edit: outstanding sounding discs, way up there in sound for GD box sets. I slightly regret not getting downloads, but my vintage system and 24/96 DAC said otherwise.