 
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
|
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
129.33.19.254
In Reply to: RE: A nasty surprise in Adobe Audition posted by Jim F. on December 15, 2009 at 10:40:39
I'm aware of what graphs show - but that's not what I personally heard on my system, for whatever reason, when playing with Audition 3.0.1 to upsample 16/44 to 24/96.
Follow Ups:
> Did your listening tests confirm superiority of "No Pre/Post filter"
> setting? . . . I'm aware of what graphs show - but that's not what
> I personally heard. . .
I didn't make any claims about the superiority of "No Pre/Post filter"
setting in Audition SRC.
I am aware that you and others prefer the filtering, while fmak does not
(he says, in http://db.audioasylum.com/cgi/m.mpl?forum=pcaudio&n=56664
"On a bright system, pre-post filtering suppresses some of the peakiness.
However, no filtering sounds better by way of sound staging, naturalness,
and open[n]ess.")
So on that basis (and, I admit, because of the graphs) I decided to
leave the filtering off.
However, the main point I was making was that I got blindsided by
Audition's installation-default behavior when reading 32-bit files.
Yes, I did, at least for 44.1 to 176.4k over many files.
44.1 to 96? Why? Unless your card doesn't work at 88.2/176.4k.
> . . .at least for 44.1 to 176.4k over many files. . .
My processing is a little unconventional. I described an earlier
version of it at:
http://db.audioasylum.com/cgi/m.mpl?forum=pcaudio&n=44031
The current sequence is similar, but simplified:
1. I start with an unadulterated (no Foobar-added dither)
16-bit 44.1 kHz CD rip. I process this in Wavelab by
eliminating DC offset and then rendering through a DirectX
chain consisting of Sony (Noise Reduction 2.0) clipped-peak
restoration (-6.1 dB level pre-reduction, post-limiter on)
followed by the click & crackle remover (don't say it! ;-> )
with "Very Conservative" preset and the rumble filter off.
This is saved as 32-bit floating point WAV.
2. SRC to 48 kHz, now using Adobe Audition. Save as 32-bit WAV.
3. In Wavelab, normalize to -2 dB, Apogee UV-22 HR dither to
24 bits, save as 24-bit WAV file.
4. Playback 24/48 file, through: 1) Meridian 518, converting
to 16 bits with Curve C noise-shaping; 2) Audio Alchemy
EDR*S chain (8 DTS Pro-32s with custom ROMs) 3) second
Meridian 518 widening 16 bits to 24 bits 4) upsampling to
96 kHz in dCS Purcell 5) process through Perpetual P-1A
"resolution enhancement" program at 24/96. Bypass the
Perpetual's SRC chip by feeding it I2S from an Assemblage D2D-1.
(I no longer use more than one P-1A and two computers in
the chain. ;-> )
5. Record the above in Sony SoundForge on a second PC.
When the entire CD is captured, process the file with
Burwen Bobcat DX (at 24/96) using the "Basic 2" setting.
Save the file as a (32-bit) WAV64. Use dbPoweramp to
convert it to a (32-bit, lossless) WavPack file.
6. Use Audition again to read in the *.WV file (using
the WavPack filter), and upsample again from 96 kHz to 192 kHz.
Save the file as a (32-bit) WavPack file.
7. Use dbPoweramp to convert the WavPack to a (32-bit) WAV64.
Read this in WaveLab 6, normalize to -3 dB, Apogee UV-22
dither to 24 bits, save the file as a 24/192 WAV64 file.
8. Use dbPoweramp one last time to losslessly compress
the file to 24/192 WavPack. Load the cue sheet in Foobar.
I'm now archiving the stages at 24/48 (WAV) and 32/96 (WavPack).
I can also use the latter to created a finished 24/96 version.
The final files can be played back on a PC in Foobar, or
used to create 24/96 DVD-V images (with the help of
Eximius DVD2One V2) or 24/192 DVD-A images (with the help
of Minnetonka Diskwelder Chrome and the SurCode MLP processor).
This is all strictly for personal use, of course!
This seems a complicated process chain. How did you arrive at it and what are the advantages?
> This seems a complicated process chain. How did you arrive
> at it and what are the advantages?
It looks more complicated than it is.
There are 3 basic components (all controversial, of course ;-> ).
1) The Audio Alchemy/Perpetual Tech "resolution enhancement"
boxes. These really just add low levels of noise.
The sonic advantages (or lack thereof) have been debated
to death here and elsewhere. Some people think they're
snake oil. YMMV. I like 'em.
I've got two flavors of these:
-- One is a non-commercial
system (EDR*S) that was demonstrated at a hi-fi show in 1996.
It uses 8 of the (commercial) DTI Pro 32 boxes (also
on the market around 1996 --
http://www.audioreview.com/cat/digital-sources/cd-players/audio-alchemy/dti-pro-32/PRD_117245_1586crx.aspx )
with custom ROMs. The non-commercial 8-box version with the EDR*S
ROMs is fixed-function: 16 bits in, and 16 bits
out (but uses 8 32-bit Texas Instruments DSPs in between).
What does it do? Well, you can think of it as re-dithering a 16-bit CD.
This system will run at 48 kHz, so I upsample from 44.1 to 48
first.
How did I get hold of it? Well, the DTI Pro 32 boxes
are available cheap on eBay and Audiogon; when I accumulated
enough of these Peter Madnick was willing to sell me the
custom ROMs, which I installed myself.
-- The second flavor of the above (really a later generation of
same thing; the code was written by the same DSP programmer) is the
Perpetual P-1A. It does something similar, but can input
and output 24 bits and run at 96 kHz.
(http://www.audioreview.com/mfr/perpetual-technologies/dacs/p-1a-/PRD_124344_2738crx.aspx )
So I upsample to 96 kHz (in real-time, using a dCS Purcell,
purchased used on Audiogon when a whole bunch of them were
being dumped) before feeding the signal to the P-1A.
(The Perpetual P-1A has a CS8420 ASRC chip in it, which is **always** in
the circuit if you feed it S/PDIF, but can be bypassed if you
feed it I2S. Fortunately, the I2S output of the Assemblage D2D-1
is compatible with the P-1A's I2S input, and the CS8420 in
the Assemblage can be defeated. Got the Assemblage used too,
of course.)
Again, some people like this "resolution enhancement", and some people
hate it (whether on principle, or because they actually hate
the sound, is hard to tell).
http://www.jeffchan.com/audio/p-1a-vs-d2d-1.html
And yeah, I've got the two "resolution enhancement" systems
in series. Maybe redundant, but I have 'em so I'm using 'em.
Is there a precedent for chaining these in this way? Well,
Peter Madnick mentioned that he used 3 DTI Pro-32s in a series
chain in his system back then, and I tried this when I
had the stock ROMs in some of my DTI Pro 32's, and liked it.
(Of course, whether any such advantage, if it exists, is
the result of the signal processing or the result of the
serial chaining of the jitter filters in the DTI Pro 32s
is anybody's guess.)
The Perpetual boxes **cannot** be easily chained, though,
because they perform a -6 dB level shift each time through.
The level can be shifted back up, of course,
with additional equipment, but that way lies madness (been
there, done that. ;-> ).
There are other minor ancillaries associated with these "Resolution
Enhancement" boxes. The S/PDIF output of the EDR*S chain
(the 8 boxes themselves are connected together via I2S)
is **extremely** unstable and hard to sync to (Madnick warned
me about this) -- the only thing I found that could sync to
it was an Apogee Big Ben. The Meridian 518s (also acquired
used) at either end of the EDR*S chain (after the Big Ben,
at the output side) do 24-> 16 and 16-> 24, respectively.
Not really necessary, of course; I could do the 24-> 16 in
the computer and dispense with the Meridian's contribution
to the "sweetening via added noise" at the other end.
2) The second (also controversial) thing I'm doing is processing
through the Burwen Bobcat software (the "DX" DirectX plugin version,
at 24/96). While "resolution enhancement" is subtle -- a matter
of ambience, "body" and "warmth" (as a timbral, not a frequency-response,
thing, if the two can be distinguished); the Burwen Bobcat is
actually doing re-equalization. It's a tone control, basically.
The "Basic 2" setting I use removes some of the "peakiness" or
"shrillness" of a lot of commercial CDs.
I may be the only person here who is still using the Bobcat software --
it got a lot of buzz for a few months (like the RealityCheck CD duplicator ;-> )
and then pretty much disappeared. John Atkinson of Stereophile
dismissed it as a tone control, not suitable for a high-resolution
playback system. Mark Levinson liked it (but was selling it bundled with
an expensive DAC). Again, YMMV.
3) And of course the third thing I'm doing is the upsampling itself
(in stages, from 44.1, to 48, to 96, and ultimately to 192). This is
also controversial, as we know. ;->
So there you have it. All it basically comes down to is:
"Resolution Enhancement" +
Burwen Bobcat +
upsampling.
Everything else (WaveLab, Audition, etc., etc.) is just means to
the end.
Another thing -- just renormalizing the audio from a CD that's been
mastered "hot" -- peaks at 0 dB or even clipped -- down 2 or 3 dB
improves the sound quality. A digital filter in a CD player
or outboard DAC will **clip**. You can see this when upsampling.
Sometimes the peak level after converting to 48 kHz
can be 2 or 3 dB higher than the peak level of the original 44.1 kHz.
It varies with program material. And of course I also do
"unclipping" as part of my initial processing. (There was a mistake
in my originally-described chain -- renormalizing at the wrong
point -- that was, very occasionally, **reintroducing** clipping.
Gotcha!)
. . . on Audio Alchemy's "resolution enhancement", in conjunction
with somebody's having heard the EDR*S system at the Stereophile
show in New York in 1996, at
http://groups.google.com/group/rec.audio.high-end/browse_thread/thread/44659620494e458e/90d03717d9b252bd
(Gabe Wiener and Steve Zipser, unfortunately, are no longer
with us. Sic transit. . .)
That has to be best example of understatement I've seen in awhile:-) Thanks for the laugh.
Jeff
wow!
My files read 32 bit in Audition and 24 + cu bit activity in Digicheck on real time playabck. The FIM wav files read 24 + cuv bits active and sound slightly different.
Post a Followup:
| FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
||||||||||||||
This post is made possible by the generous support of people like you and our sponsors:
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2025, Audio Asylum, all rights reserved ]
