avatar

reynaud

Silverado

Posts: 196 Member Since:13/07/2011

#41 [url]

Mar 11 14 3:15 AM

mdm wrote:
how much memory/data does 60 minutes of 2 track audio amount to?


A source with a duration of roughly 10 minutes has the following file sizes in the different formats (all stereo):
DXD (24bit/352.8kHz): 1.12GB
BWAV (24bit/96kHz): 143MB
DSD64 (2.82MHz): 272MB
DSD128 (5.64MHz): 579MB

Surround will obviously more than double the various file sizes. File sizes also vary between converting between DXD and DSD (and vice versa) and recording natively at that resolution.

cheers,
Reynaud

 

Quote    Reply   
avatar

reynaud

Silverado

Posts: 196 Member Since:13/07/2011

#42 [url]

Mar 11 14 3:34 AM

dcollins wrote:
As I see talk of 384k, 5.6MHz, etc.

At at guess, you won't be interested in what some call OctupleDSD (DSD512 / 22.5792MHz). Apparently the JRiver media player (and a few others) support the format and a few nondescript DACs. I can already feel your eyes rolling.

I tend to think DSD256 may possibly be the new marketing rate in much the same way Digidesign was marketing 192kHz as the next breakthrough. Although, DSD256 softens the hi-end slightly more when compared to DSD128 (maybe to the same extent as between DSD64 and DSD128 if not more so) and does at first listen appear more euphonic.

cheers,
Reynaud

Quote    Reply   
avatar

Mike Rivers

Aqua Marine

Posts: 2,503 Member Since:13/10/2012

#43 [url]

Mar 11 14 5:14 AM

reynaud wrote:
I tend to think DSD256 may possibly be the new marketing rate in much the same way Digidesign was marketing 192kHz as the next breakthrough.

 
"Marketing rate!" I love it. We used to talk about "marketing bits" and as far as I know, we're still not truly at 24-bit resolution of anything but noise.



For a good time, call mikeriversaudio.wordpress.com

Quote    Reply   
avatar

john bailey

Silverado

Posts: 126 Member Since:06/02/2011

#44 [url]

Mar 11 14 3:33 PM

tom eaton wrote:
I have read people complaining that a realtime trip a/d to d/a is more transparent than a/d to hard drive, playback to d/a.  I don't know if that's clocking (which could make perfect sense to me) or what, and I can't think of a way to slave a tape machine to a  computer or vice versa that doesn't put one or the other at a disadvantage, but I wonder about that test.  I think Paul is brilliant, don't get me wrong... but I don't think a realtime trip a/d/a is the whole answer to transparent REPRODUCTION.

t

 

I would proudly include myself in the group of people complaining that a trip through A/D then D/A is more transparent than A/D to LPCM WAV File.  And I can prove it.

What you're hearing while a mix is running 'live' on your DAW, is not fully represented in the WAV file that is digitally captured.  Likewise, if you're doing a tracking session, what you're hearing while the band is out on the floor playing, vs what you all hear when they come in for a playback is a disappointment (times x tracks in record).  Again, I can prove it to you if you like.

Bottom line is...  If you actually think that the digital process for recording live digital streams to LPCM WAV (or AIFF, or whatever) is accurate and complete, then I present a challenge.  If somebody will volunteer their Pryamix / DSD (Horus?) system, I will gladly fly to your city and prove it to you.  If I win, then you cover my flight.  If I lose, then I pay for your studio time.

Digital audio is still very much in it's infancy....
Cheers, eh?

Quote    Reply   
avatar

cgc

Gold Finger

Posts: 944 Member Since:06/02/2011

#45 [url]

Mar 11 14 6:30 PM

Can you give a rough outline of how the code for writing PCM to disk would degrade the sound compared to the stream coming off the ADC? And how a DAW would transparently move the data from ADC to DAC but not write it to a filesystem properly.

I'm at a total loss as to how that would occur and need to see some serious hardcore proof of this. Not the usual 'use your ears' bullshit.

Quote    Reply   
avatar

podgorny

Platinum Blonde

Posts: 2,328 Member Since:27/01/2011

#46 [url]

Mar 11 14 7:13 PM

cgc wrote:

Not the usual 'use your ears' bullshit.



You're just bitter because you lack the aural acuity to discern the subtle differences between two totally identical things. 

Quote    Reply   
avatar

john bailey

Silverado

Posts: 126 Member Since:06/02/2011

#48 [url]

Mar 12 14 10:26 AM

Yeah, I can't stand the audiophile BS talk...

We can do this with your speakers off.  I'm purely interested in the science, and I'd love to have the most adamant naysayers present, if possible!

Offer still stands!

Cheers, eh?

Quote    Reply   
avatar

Mike Rivers

Aqua Marine

Posts: 2,503 Member Since:13/10/2012

#51 [url]

Mar 12 14 1:23 PM

john bailey wrote:

We can do this with your speakers off.  I'm purely interested in the science, and I'd love to have the most adamant naysayers present, if possible!

Since your demonstration doesn't depend on listening, why not explain your procedure so we can either try it at home or poke holes in it, or both?



For a good time, call mikeriversaudio.wordpress.com

Quote    Reply   
avatar

john bailey

Silverado

Posts: 126 Member Since:06/02/2011

#52 [url]

Mar 12 14 3:51 PM

mikerivers wrote:
Since your demonstration doesn't depend on listening, why not explain your procedure so we can either try it at home or poke holes in it, or both?
 

OK.

To demonstrate, here are the conditions and the test equipment.

For the test equipment, you'll need:

1.  A DAW, of your choice at 44.1kHz, feeding a D/A converter.

2. A DSD recorder, or even better, a DSD workstation (you'll need that later)

3. A Common clock, at 44.1kHz (seems like we're stuck with that, but it'll do)

4. A DSD workstation, that will allow you to invert and compare files

For the control part of the test:

1. Play a stereo WAV file out of the D/A converter and record it into the DSD Recorder (or DSD Workstation).

2. Play the exact same stereo WAV file out of the D/A converter, and again, record it into the DSD Recorder (or DSD Workstation)

3. Line up these two files in the DSD Workstation (inverting one of them) and create a null.  You should be left with the sum of two passes of residual noise
*** for anyone attempting to perform the same test by downconverting the DSD files to PCM in order to compare, you will notice that you can't get a perfect null.  This is because the 44.1kHz external clock feeding the DSD recorder does NOT always refer to the exact same sample position - at DSD128 there are obviously 128 sample for every clock cycle)

4.  Do not proceed until you can create an audio null with the control test.

5.  Open up a complex mix session on your source DAW, that hopefully employs lots of live, random elements - reverbs with long decay, delays, modulation effects, etc...

6. Play this mix session out of your source DAW and record it into the DSD recorder, while simultaneously recording the output that feeds the D/A converter to a LPCM WAV File.

7. Play the WAV file that was captured in the source DAW out of the D/A converters (at exactly the same level) and record it into the DSD Recorder (or DSD Workstation).

8. Again, play the WAV file that was captured in the source DAW out of the D/A converters (at exactly the same level) and record it into the DSD Recorder (or DSD Workstation).

9. You now have three stereo DSD files for the test.  #1 is the mix session running 'live' feeding the D/A converter, and #2 and #3 are both playback passes of the mix, captured as LPCM in the source DAW.

10. Create a null, using pass #2 and pass #3.  (you've done this before)

11.  Now, attempt to create a null, using pass #1, and report your findings to the group!
 

Quote    Reply   
avatar

Mike Rivers

Aqua Marine

Posts: 2,503 Member Since:13/10/2012

#53 [url]

Mar 12 14 4:55 PM

john bailey wrote:


To demonstrate, here are the conditions and the test equipment.

For the test equipment, you'll need:

1.  A DAW, of your choice at 44.1kHz, feeding a D/A converter.

2. A DSD recorder, or even better, a DSD workstation (you'll need that later)

3. A Common clock, at 44.1kHz (seems like we're stuck with that, but it'll do)

4. A DSD workstation, that will allow you to invert and compare files

 

Ah! #4 is the gotcha. You probably have a Pyramix or Sonoma, and I'm not even sure if they can invert without converting to PCM and back again.

But basically what you're demonstrating is that no clocking system is perfectly repeatable. I suspect that anyone who really cares already knows that. So, at this point, unless I didn't completely follow your experiment, I'll just say "So what?" It's what we have to live with since we can't hear live music all the time. 



For a good time, call mikeriversaudio.wordpress.com

Quote    Reply   
avatar

cgc

Gold Finger

Posts: 944 Member Since:06/02/2011

#54 [url]

Mar 14 14 12:02 AM

podgorny wrote:

cgc wrote:

Not the usual 'use your ears' bullshit.



You're just bitter because you lack the aural acuity to discern the subtle differences between two totally identical things. 

That or I know turning my head 10 degrees is far greater a sonic shift than what is being discussed here.

Quote    Reply   
avatar

zenmastering

Silverado

Posts: 155 Member Since:21/02/2011

#57 [url]

Mar 15 14 2:36 AM

Pyramix converts DSD256 to 32fs / 32bit PCM (aka "DXD4") for the duration of fades or any clip with a level change applied. This doesn't happen in real-time, only in the 'DSD-Render' mode, which essentially copies the source DSD256 files into a new output file until such a fade or level change occurs and then remodulates those to DSD256.

DSD128 converts to 16fs / 32bit for the same process.

There is an AES paper, from McGill University & Philips Research, promoting the use of high sample rates for the initial A/D, even if the final output is much lower resolution. I've done similar, empirical tests, well before this paper was done (and since): I completely agree.

One sec...

AES31-000069
"Which of the two digital audio systems best matches the quality of the analogue system?"

Graemme

Quote    Reply   
avatar

john bailey

Silverado

Posts: 126 Member Since:06/02/2011

#58 [url]

Mar 15 14 8:27 PM

zenmastering wrote:
There is an AES paper, from McGill University & Philips Research, promoting the use of high sample rates for the initial A/D, even if the final output is much lower resolution. I've done similar, empirical tests, well before this paper was done (and since): I completely agree.

One sec...

AES31-000069
"Which of the two digital audio systems best matches the quality of the analogue system?"

Graemme
 

Graemme, yes, that has been my experience.  In my tests here, the differences are quite noticeable, even when both are converted all the way down to 128kbps MP3.

I presented an e-Brief at AES131 on this phenomenon...  https://secure.aes.org/forum/pubs/ebriefs/?ID=207 
(or if you're not a member...  https://dl.dropboxusercontent.com/u/59174624/AES_131_e-brief_BAILEY_Playback%20Disappointment.pdf)

Harold (when he was at CIRMMT / McGill) and Doyuen Ko were kind enough to help me run some test files on this...

The difference here, though is that I firmly believe  (and will prove to anyone willing) that the 'loss' is happening when the live digital stream is captured / stored / retrieved as a WAV file.
 

Quote    Reply   
avatar

neworleanssteve

Tin Man

Posts: 39 Member Since:06/02/2011

#59 [url]

Mar 16 14 2:54 PM

zenmastering wrote:


There is an AES paper, from McGill University & Philips Research, promoting the use of high sample rates for the initial A/D, even if the final output is much lower resolution. I've done similar, empirical tests, well before this paper was done (and since): I completely agree.

One sec...

AES31-000069
"Which of the two digital audio systems best matches the quality of the analogue system?"

Graemme

YES ! "Free Willy" looks so good in VHS, because they shot it in Panavision.

Thats what me mentor told me... way back when "Free Willy" was just comming out on VHS
Steve Daffner 
www.frenchmenstreetrecords.com

Quote    Reply   
avatar

john bailey

Silverado

Posts: 126 Member Since:06/02/2011

#60 [url]

Mar 19 14 12:49 PM

mikerivers wrote:
Ah! #4 is the gotcha. You probably have a Pyramix or Sonoma, and I'm not even sure if they can invert without converting to PCM and back again.

But basically what you're demonstrating is that no clocking system is perfectly repeatable. I suspect that anyone who really cares already knows that. So, at this point, unless I didn't completely follow your experiment, I'll just say "So what?" It's what we have to live with since we can't hear live music all the time. 

Mike, I actually don't, but really wish I did - especially with all the turmoil at Avid these days.. 

No, we're not talking about clocking here...  The point is, that in order to demonstrate the flaws inherent in a given system, you must use a much more resolute system to measure and demonstrate them.

To be honest, I'm a bit discouraged that people seem to be resigned to the 'good enough' mentality.  Entirely understandable, considering the current climate of the business, but I have a hard time giving in...

Still no takers, eh?  I'd really love to do this...

Cheers, eh?
 

Quote    Reply   
Add Reply

Quick Reply

bbcode help