 
|
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 |
157.150.192.237
I've had a weekend to play with this, but I've listened to
quite a variety of stuff.
I never would have believed that an SRC could make **this**
much difference, and this is just 44.1->48.
I've been listening through the "front end" of my signal-processing
chain, in real time, via a computer with modest Cambridge
Soundworks satellite-sub speakers (very near-field, almost
headphones) and a not-so-modest Cary Cinema 11 D/A converter
(this computer doubles as my TV, not that I have time to watch
TV).
Source files were pre-processed via Sony Noise Reduction 2
DX plug-in in WaveLab (clipped peak restoration, -6.1 dB
pre-attenuation, post-limiter on; followed by Click and
Crackle removal, "Very Conservative" preset, rumble filter
off), processed and saved as 32-bit float. These were played
back in Foobar 0.8.3 as 32-bit files. Baseline SRC
to 48 kHz using Secret Rabbit Code, "best sinc", slow-mode on;
final Foobar output dithered to 24 bits with "Strong ATH
noise shaping". 24/44.1 Toslink S/PDIF out from sound card (Audiotrak
Prodigy 7.1, ASIO driver) to an RME ADI-192DD, thence to a
Meridian 518 set to reduce 24->16 bits via "Curve C" noise
shaping, thence to the Audio Alchemy EDR*S chain, thence
to an Apogee Big Ben, thence to second Meridian 518 set to
expand 16->24 bits (by filling the lower 8 bits with noise, also
"Curve C"), thence to a dCS Purcell doing SRC from 48 to 96
(noise shaping set to "Auto"), thence to an Assemblage
D2D-1 in "Transparent" mode converting S/PDIF to I2S, thence
to a Perpetual P-1A set to run its "Resolution Enhancement"
program at 24/96, and finally to the Cary Cinema 11 at 24/96.
Analog output of the Cary goes to a Musical Fidelity X-10v3
tube buffer, and then to the Cambridge Soundworks amplifier/woofer
box.
All right, so the comparison here was going from (pretty
highly regarded) Foobar 0.8.3 Secret Rabbit Code (best sinc,
slow mode) to playing the same files processed by Saracon
(input files are 32-bit float from the Sony pre-processing; output
files from Saracon are 32-bit, files are post-processed in Wavelab
to normalize to 0 dB and then dither to 24 bits using Wavelab's
Apogee UV-22HR. The Saracon-processed files were also played
back in Foobar, through the chain described above.
Note that one difference between the Secret Rabbit and
Saracon scenarios is that the (32-bit/44.1 kHz) files I was
playing back and upsampling in real time with Secret Rabbit
had a peak level at -6 dB, give or take, whereas the files
offline-processed by Saracon had then been normalized back
up to 0 dB. So the Secret Rabbit might have had a 1-bit
disadvantage at the point where the stream gets converted from
24 bits to 16 bits by the first Meridian 518 (the reason for
this, once again, is that EDR*S is strictly 16 bits in and 16 bits
out).
OK, so what's the verdict?
**Huge** difference (as audiophiles use the word "huge" ;-> ).
I've used both flavors of Secret Rabbit; more recently
I've used r8brain Pro and SoX and Audition 3. I can't
say I've heard any great differences among these.
Saracon is at a whole different level.
The differences are the sorts of things one would
typically ascribe to a better-resolving D/A converter,
but remember this is **all** the result of digital
filtering, and increasing the sampling rate only
from 44.1 to 48.
On "difficult" recordings, there's a layer of upper-
midrange glare removed (so much so, that in my
back-end processing I've resolved to use Burwen Bobcat's
"lightest touch" Basic 3 setting, rather than the
darker Basic 2).
The high frequencies (or what I **perceive** as the
high frequencies at my age) are exquisite and highly
resolved. You know the routine. Cymbals sound more
like cymbals and less like noise. Other percussion,
like orchestral bells, are highly resolved. There
are layers more resolution -- buried details are
more audible, or audible for the first time.
Even speech is more intelligible.
Really **really** difficult recordings shine, such as,
for example, a torture test that nothing before
has really passed -- an old MCA double CD entitled
"The Extraordinary Roger Voisin", from an even
older analog recording on the Kapp label of baroque
trumpet recorded **really close up**
http://www.amazon.com/Extraordinary-Roger-Voisin-Manfredini-Altenburg/dp/B000009I77
The timbre of Voisin's trumpet is totally screwed
up (to the point of sounding like gross clipping) through
garden-variety CD players, and even through better-
pedigreed equipment. Saracon conversion made this sound
pleasant for the first time -- I could listen to the
whole thing without having to change the record.
Difficult pop recordings like, oh, Annie Lenox's
"Medusa" or Aretha Franklin's "Who's Zoomin' Who"
are reproduced with considerably less glare and
congestion. In all genres, the presentation is
less congested and more relaxed. Listening fatigue
is reduced.
Anyway, this was $850 better spent, IMO, than most other
audio purchases I've made.
Saracon is simple to use (though it requires a USB dongle --
as does WaveLab 6; my USB ports are getting filled up
with these things!). You can add files to a batch-conversion
list and let it grind away overnight. It computes
at 64 bits internally, and uses POW-R dither to convert
to whatever output bit depth you choose. It will output
64-bit floating or 32-bit fixed, though there aren't
many DAWs that can use these -- I think Cakewalk Sonar 8.5
can read 64-bit floating-point WAV files. You can
tell Saracon to apply gain reduction to an input
file before applying the SRC. It won't, unfortunately,
do more sophisticated DAW-ish things like normalizing
a file to a given peak level following SRC.
It will support Sony's WAV64 format for files bigger
than 2 GB (necessary if you're going to process entire
albums at 96 or 192 kHz without going to the trouble
of splitting them up).
So there you have it. There may be equally good SRC software
out there -- iZotope 64-bit SRC also has a good reputation.
Offline upsampling isn't for everybody, of course.
It's at the extreme edge of practicality, if it's practical
at all -- it's extremely labor intensive, and life is
short.
Follow Ups:
I use a Korg DSD recorder to RIP vinyl. Any benefit to using saracon to convert DSD to PCM? Or even to process PCM recordings from the Korg?
Thanks for posting this. I am firmly in the camp that is for DSP "home remastering". I have done some offline upsampling of hard sounding CD's simply by saving them in Wavelab at a higher sampling rate. It did smooth out things a little, but was not a dramatic improvement. I tried Rbrain and could hear no improvement at all. I don't think I'll spring for $850 for Weiss Saracon but Izotope seems tempting.
If you listen to classical recordings you might take a look at a free spacial enhancement VST plugin that is supposed to be similar to a respected $1000.00 professional mastering plugin: K-stereo Ambiance Processor. I have played around with it. For me there was and is a learning curve, but it seems to be very elegant and natural.
Jesus effin Christ. Your post there is why folks in other forums are not in a hurry to get anywhere near a computer for audio.
You did all that, did you actually listen to any music?
You know what I do, I grab beer, lately a New Belgium 1554, select the proper inputon the pre-amp, sit down on the couch and press play.
You have seriously made this about 15 billion times more complicated than it needs to be. My god.
Happiness is a clean record, and warm tubes!
> You did all that, did you actually listen to any music?
Actually, I'm listening to music all the time I'm doing
that.
;->
Some audiophiles like to attempt to improve sound by spending a ton of money on hardware. That's what appeals to them the most. Others appreciate the potential of DSP and like playing around with software. Isn't it more sensible for Jim F. to try to get a better sound by investing in an $850 professional program than by spending $10,000 on an EL34 tube amp? To me getting the sound you want purely by changing hardware is like changing light bulbs every time you want to change the brightness in the room. Why not just get a dimmer switch?
It is more sensible to expend ones effort on room acoustics and treatment than anything else audio playback related. And most audiophiles don't do it. They spend more money on gear, and these days lots of effort on computer software settings. That's crazy.
With Jim's DSP chain he could probably simulate the distortion profile of any tube amp you want quite nicely! he he he
And this from a guy who loves 6L6 and 300B tubes...
Cheers,
Presto
> With Jim's DSP chain he could probably simulate the
> distortion profile of any tube amp you want quite nicely!
Well, no, it's not quite that -- parametric. ;->
I was kind of amused to discover, however, in conjunction with
my research into an LP digitizing project that I may not now
get to in this lifetime (I've had a Versa Dynamics turntable
staring at me accusingly for the past 4 years now :-/ )
that Diamond Cut Productions' DC7 "virtual preamp" software
(for doing RIAA correction on vinyl digitized through a
"flat" preamp -- like the Instruments input on the Prism Orpheus)
will color the sound for you according to the distortion
characteristics of a whole list of tubes. They have a setting
called "high-end tube preamp" or something like that. ;->
"Why not just get a dimmer switch?"
Or some tone controls. I agree that the process describe here is prohibitively complex. And I seriously doubt it yields better results than a bit of eq. There is no doubt that audio tweaking, whether hardware or software-based, is a hobby in and of itself, apart from enjoying music well-reproduced.
P
> I seriously doubt it yields better results than a bit of eq.
Ah yes, the days of the ubiquitous parametric equalizer.
I remember them. Solve all your audio problems with this
cheap little Metrotec box!
;->
Nah. Much too complicated. A well-calibrated tone knob should do the trick.
p
> Nah. Much too complicated. A well-calibrated tone knob
> should do the trick.
Can't you just imagine J. Gordon Holt (of blessed memory), in a
mischievous mood, advising his readers "Don't waste your money
on those hifalutin' Koetsu and Kiseki and such-like exotic
cartridges whose coils are wound from mithril spun by the Elves
of Lothlorien to a thinness finer than Elf-hair. Just buy a
Shure and use your equalizer." ;->
Later on, of course, there were "equalizers" like the
Cello Audio Palette, and in the computer world of nowadays
there's Burwen Audio Splendor. Could it be done --
making a Shure sound like a Koetsu Rosewood Platinum Signature?
Maybe. Who knows?
Remember the Bob Carver Amplifier Challenge? He swore
that he could "voice" one of his solid-state t-amps
(t for "tube", get it?) to sound exactly like any
exotic tube amplifier the Stereophile
reviewers cared to name. That didn't turn out too well,
as I recall. But Carver did up making an exotic tube amp
of his own -- the Silver Seven. How big is **your**
circuit breaker?
http://community.klipsch.com/forums/storage/4/768823/Carver%20Silver%20Seven_Spain1.jpg
Audio tweaking begins from the enjoyment of music seeking to have it better reproduced and to give the tweaker an active role in it. It can however take on a life of its own to the point of putting the music itself and even fidelity second (but not "aside"). When this happens, tweaks take on an idiosyncratic and inordinate value in relation to simply reproducing the music well.
> Audio tweaking begins from the enjoyment of music seeking
> to have it better reproduced and to give the tweaker an
> active role in it. It can however take on a life of its own
> to the point of putting the music itself and even fidelity
> second (but not "aside"). When this happens, tweaks take
> on an idiosyncratic and inordinate value in relation to
> simply reproducing the music well.
Perhaps. And it's true of system "upgrading" too (i.e.,
buying new equipment), not just "tweaking" (pounding furniture
glides into the floor and propping your speakers and amplifiers
on them, putting Webster's Unabridged -- poor man's Shun Mook --
on top of your CD player to "damp" it, etc.)
What happens though, is what B. F. Skinner called a "variable-ratio
schedule of reinforcement". Sometimes, you get a real jolt
of the sweet (dopaminergic neurons squirting out neurotransmitter ;-> )
as a result of something you've done. I can still remember
the first time I heard Audio Research tube gear back in 1976
(particularly the SP-1-a preamp) -- what a rush! Other things,
fairly few and far between, have given me a similar thrill
over the years. Yes, putting my Quads on furniture glides
tightened up the bass very satisfyingly (no, I don't have
spikes on the bottom and sand inside the Arcici stands these
days -- I use a separate powered woofer instead. ;-> ).
A big project I did back in -- '94? '95 -- involving a
Radio Shack Optimus CD-4300 battery-powered CD transport
feeding a pair of battery-powered DTIs feeding a Theta
DS Pro Prime with its DAC chip also battery powered gave
me quite a thrill. And hearing Eximius DVD2One software CD
upsampling was another unexpected thrill (By contrast,
cheap ASRC-based upsampling did not give me the sensations
Jonathan Scull experienced reviewing the dCS 972, though
my first such device -- a Bel Canto DAC-2 -- was very nice
indeed, though probably, I now realize, for other reasons
than simply the upsampling. Later instances of the same
thing -- "I've got 24/96, now give me 24/192!" -- were not as
edifying.)
And so it goes. Other things, while eagerly anticipated to
the point of frenzy, like the Compact Disc itself, turned into an
almost heart-breaking disappointment when they finally materialized
(and thereby hangs a 25-year-long tale).
And that brings us to the second source of the problem.
Consumer audio reproduction has always been poised right
at the threshold of the "barely acceptable". It could almost
have been designed to lure susceptible people into a
kind of gambling behavior by providing just the occasional
tantalizing glimpse of paradise[*]. (It could almost have been
done deliberately, but I don't believe in conspiracy
theories -- it's just the way things work, what with the cost
and technical difficulty of superior audio reproduction,
the cost-constrained nature of consumer manufacturing, and
the need for economies of scale to keep costs down --
manufacturers need to target the mass market, and the
mass market **just doesn't care**.) That's part of the
heartbreak of digital audio -- it was supposed to be **better**
than adequate by a substantial margin (compared to
mechanical tonearms and cartridges and analog grooves
pressed in vinyl) -- it certainly looked like it was going
to be, on prima facie evidence (whooped up by both
marketers and "true believing" engineers). And then it
turned out to be **worse**!! And nobody in the "responsible"
press -- only the "crazies" at The Absolute Sound --
would even admit it! Talk about Charlie Brown and the football.
So, digital has turned out to be every bit as difficult to
get right as analog ever was.
And the beat goes on.
[*] Talk about "glimpses of paradise". Long before there was a buzz
about RFI and power-line interference, back in the mid-70's,
I was continually frustrated by the fact that, when I listened
to my (LP) system before going to work, it sounded **great**.
I'd anticipate returning to it all day, and then when I got
home at 5:30 and turned it on again, it would sound like **crap**.
I always thought this was a psychological phenomenon (and maybe
it was, but I don't think so -- in a similar way, I independently
discovered 1) inner-groove distortion and 2) listener fatigue,
before I ever heard the terms). I now realize that this was
probably because power is dirtiest in the late afternoon, when
all the household appliances are on. I now have all my systems
on power conditioning (PS Audio Power Plants) and, sure enough,
I get consistent sound.
> . . .(particularly the SP-1-a preamp). . .
I meant SP-3a (or SP-3a-1), for those of you with
long memories. ;->
I bought the Audio Research SP 2c back in the day. Nothing will ever seem to live up to the retrospective perfection. I had a system with that and its mate the Dual 50 F-1a 50w/ch tube amp feeding Magnaplanar Tympani 1-U three panel per channel electromagnetic screen speakers. In fact I have these still stored in my basement though probably unusable now. I just could not part with them.
I like your account of how things go. I was in particular trying to correct a bit what PP said. I don't think it is always the case that tweaking and upgrading take precedence over appreciating the music as well reproduced as one can. I sure try to keep it in its place, but like everyone I get seduced now and then.
> I had a system with that and its mate the Dual 50
> F-1a 50w/ch tube amp feeding Magnaplanar Tympani 1-U
I had, along with the SP-3a-1 (bought new for, what was it,
$795 ? -- anyway, a staggering sum of money for a preamp in
those days) a Dual 75 (bought used). And a pair of
Magnepan MG-IIs.
Years later, I sold the MG-IIs to a friend, and bought
a pair of Tympani IVs. With the ribbon tweeters. Total
overkill! (Buying them was a nostalgia trip. Magneplanar
Tympani (presumably the 1U, like you had) were
my **very first** introduction to "high-end audio" back
in 1973.) I **never** had the space to set those up
properly. They **literally** divided the rooms I had them
in, and overdrove the acoustic space pretty badly.
Had fights with the upstairs neighbor over them, more
than once (not that I ever played them that loudly; they
were just overpowering the small room **and** the neighbors).
So I finally traded them in and bought the Quad ESL-63s
(used). I still have and use the Quads, but they threatened
to be the death of me, twice. Another trip down memory
lane: I knew about the Quads' reputation for finickiness,
going back to the original "ELS-57" firescreens, but
some reviews in Stereophile indicated that the ESL-63s
were perfectly happy with high-quality solid state
amplification. So I bought a Krell KSA-80. Can't get
any higher solid state than that, right? Of course,
I was fronting all this with early-90s digital, through
a modest solid-state line stage (an Adcom GFP-565, which
I also still have and have uses for). Well, after
the initial love affair with the Quads wore off, I started
being plagued by some kind of high-frequency irritation.
Not just the typical digital upper-midrange glare,
but something on the threshold of hearing, like what
Todd Krieger says he hears with upsampling. Like very
high-pitched noise. I thought I was getting tinnitus!
I started going nuts swapping equipment, playing with
cables, etc. I finally had to ditch the (expensive,
even though used) Krells and go to tube amplification
(a pair of VTL Tiny Triodes at first, then a Counterpoint
SA-4 OTL tube amp purchased used -- a major reliability
headache, but that's another story). Before things
settled down with the Quads (and they never **really**
settled down) I was actually in tears over the sound
of that system (**the** system, my only system, at the
time). So that was my first pound of flesh extracted by
the Quads.
The second pound of flesh had to do with the bass.
Quad ESL-63s (unless they're rebuilt by Crosby Audio Works,
or some such third-party outfit) have a structural resonance
around 50-60 Hz that can be excited by just the right
program material. It's the famous Quad rattle. It
doesn't have to be playing loud, and it isn't always
the kind of program you'd expect to hear it on (like
an Enya CD), it can happen, unpredictably, with
almost anything. Very disconcerting, and more and more
irritating, like rubbing the same sore spot,
the longer you have the speakers. I was finally
ready to ditch them 10 or 12 years ago, but at the
same time I **hated** the idea of parting with them, so I decided
(riskily) to throw more money at the problem by getting
a Velodyne HGS-12 woofer to get the bass out of
the Quads. Well, the crossover in the Velodyne wasn't
nearly steep enough, so I had to supplement it with
a Bryston 10B SUB, crossed over at 100 Hz and 18 db/octave.
That took care of about 80% of the problem. But later
I replaced the Bryston with a Marchand XM126-2AA
two-way tube crossover, 24 db/octave and fixed 100 Hz
crossover frequency. **That** finally took care of
the problem, 100%.
My latest discovery about the Quads is that they **love**
digital amps. Who'd have believed it! So now I'm
very green -- I drive the Quads with a modest Little Dot T-150
Tripath amp, and I've retired the Jadis JA-80s and the
AtmaSphere M60 MkII OTLs, and the speakers seem just
as happy as they can be. With a conrad-johnson PV12-L
line stage (ca. 1994) before the Marchand crossover,
and currently sourced by the Cary 306 SACD used as
a D/A converter. It's a good thing the Quads so graciously
allowed me to go green. The previous AtmaSphere amps
(8 triode output tubes and 4 driver tubes per side),
each on its own PS Audio P-1000 power conditioner,
would blow the room's circuit breaker after about
an hour or so of listening, even with all the lights
off.
"I don't think it is always the case that tweaking and upgrading take precedence over appreciating the music "
-- I don't think it is always the case either.
p
In fact, I have learned from hard earned experience to stop tweaking once something good is obtained. Further tweaking can make things worse.
I therefore have two systems; one for experimentation and one for listening. The experimental system helps to upgrade my main system if something works.
avid audiophiles are not passionate music lovers. I bought my amp used from an audiophile with a $50K system who barely had 30 CD's. In this forum you probably find the least reference to music. One man's hobby is another man's quirky obsession. If your hobbyist goal is to manipulate electrons until they yield the sound you want, DSP is as sensible as endless equipment swapping. A focus on software is just as valid, and more economical, as focusing on hardware.
This is all we use... especially for all our DSD/DXD work.
> This is all we use... especially for all our DSD/DXD work.
Yes, of course, that (the DSD version) is a more expensive
edition of the product.
The one I have is PCM-to-PCM only.
Why are you converting the sample rate?
P
P,
I'm no expert on this, but, I believe the technical explanation of possible value is in providing an 'easier' signal for some DACs and their filters to deal with, and thereby reduce the deleterious effects of same.
clay
> I'm no expert on this, but, I believe the technical explanation
> of possible value is in providing an 'easier' signal for
> some DACs and their filters to deal with, and thereby reduce
> the deleterious effects of same.
Yep, I also think that's the technical explanation for why upsampling
"works" (if you think it does!).
And not just "some" DACs, either, at this stage of technology.
The real-time DSPs in DAC chips are still heavily cost
constrained. They simply cannot do what Weiss Saracon (or
for that matter, "iZotope 64-bit SRC") is doing.
(Note that the original ASRC chips used for "on the cheap"
upsampling following on the buzz created by the dCS 972 in 1999
were **just another** flavor of cost-constrained chips!
The newer Burr-Brown SRC4912 and Analog AD1896 are supposed to
be rather better.)
Maybe in another decade or two Sony or Panasonic will be
able to pop that much computer power in an ordinary CD,
or DVD (or maybe only Blu-Ray and beyond) player.
Still, some (expensive) players and DACs have had
custom-programmed DSPs in them even as far back as
the late 80s (Theta DSPre, "Digital Done Right" ;-> ).
Take a look at the technical specs of the original
Philips SAA-7030 4x oversampling digital filter (used
in their very first CD-100 CD player -- Magnavox FD-1000
in the US) for a good shudder. (12-bit filter coefficients,
IIRC, and truncation -- no dither.)
http://www.marantzphilips.nl/the_evolution_of_dac_the_digital_filter/
http://www.marantzphilips.nl/img/info/philips/philips-cd100/philips-cd100-3.jpg
BTW, Jonathan Scull's review of the dCS 972/Elgar upsampling
combo back in 1999 remains absolutely the best piece of audio
porn I've read in my entire life. Boy, did that have my juices
flowing ten years ago!
"I felt that somehow, from deep in the soup of high-speed
floating-point calculations, the true nature of the music
emerged completely unscathed. I reveled in 24/192's apparent
felicity to timbre, timing, pace, its naturally quick
rise time, the highly detailed leading edge and follow-on
bubble of acoustic information rife with detail. Acoustic
bloom lingered in the air before twinkling out into the
ambient noise floor with the grace and naturalness of
an Olympic diver."
http://www.stereophile.com/digitalprocessors/260/
(He sounds a little like the late J. G. Ballard, here.)
What do you suppose is the most important part of a Monk solo? The notes or the silences?
p
The pianist Artur Schnabel reportedly said:
"the notes I play no better than many pianists, but,
the pauses between the notes, ahh, that is where the art resides"
clay
> What do you suppose is the most important part
> of a Monk solo? The notes or the silences?
Is that a Zen riddle? What is the sound of one hand
clapping?
"Love and marriage, love and marriage,
go together like a horse and carriage. . .
Let me tell you, brother,
you can't have one without the other."
Jim, I agree with your observations. I use Saracon to upsample my music to 24/96KHz. 96K is the native frequency of my Tact amp, and any frequency other than 96 fed to it will be sample rate converted. My experience is that feeding it with 96 gives better sound, probably because it avoids the cheap-chip version of what Saracon does via software.
“I never would have believed that an SRC could make **this**
much difference, and this is just 44.1-> 48.”
A 44.1-> 48 conversion at best wouldn’t sound all that great………. This is exactly what Microsoft kmixer does with CDs and CD-quality files, which is a process a lot computer audiophiles seek to bypass or eliminate. Better leave the signal alone at 44.1, if possible.......
“Source files were pre-processed via Sony Noise Reduction 2
DX plug-in in WaveLab (clipped peak restoration, -6.1 dB
pre-attenuation, post-limiter on; followed by Click and
Crackle removal, ‘Very Conservative’ preset, rumble filter
off), processed and saved as 32-bit float.”
What recordings is this being applied to? It looks like you’re trying to “enhance” the mix, which might work in some cases, but I’m afraid would make things worse in most others…… The “click and crackle removal” suggests you’re sampling the music off vinyl.
32-bit float..... At what sample rate? (An A/D converter would need more than 32 bits to filter the signal properly.)
“These were played
back in Foobar 0.8.3 as 32-bit files. Baseline SRC
to 48 kHz using Secret Rabbit Code, ‘best sinc’, slow-mode on;
final Foobar output dithered to 24 bits with ‘Strong ATH
noise shaping’.”
Instead of processing your source signal to 32 bits and then later convert it to 24/48, why not just process your source signal to 24/48 in the first place? You would no longer need the SRC, and you’d also save disk space…………
“24/44.1 Toslink S/PDIF out from sound card (Audiotrak
Prodigy 7.1, ASIO driver) to an RME ADI-192DD, thence to a
Meridian 518 set to reduce 24-> 16 bits via ‘Curve C’ noise
shaping, thence to the Audio Alchemy EDR*S chain, thence
to an Apogee Big Ben, thence to second Meridian 518 set to
expand 16-> 24 bits (by filling the lower 8 bits with noise, also
‘Curve C’), thence to a dCS Purcell doing SRC from 48 to 96
(noise shaping set to ‘Auto’), thence to an Assemblage
D2D-1 in ‘Transparent’ mode converting S/PDIF to I2S, thence
to a Perpetual P-1A set to run its ‘Resolution Enhancement’
program at 24/96, and finally to the Cary Cinema 11 at 24/96.”
This is insane……… There are so many conversions (I counted five), it makes me cringe…….
You’re in essence taking a 32-bit signal, downconverting it a few times to 16/44.1, and then upsampling it a couple times back to 24/96. (I once likened this to taking fresh squeezed orange juice, and then adding water, aspartame, and artificial flavors.)
I guess if you like how it sounds, any technical criticism shouldn’t matter…… But then again, I'm afraid you might not realize what you’re hearing…………
All you need to do is sample your original source signal (the very first step) at 24/96 (if it’s analog), and play it back at the native rate. No conversions necessary. The playback will be a *lot* cleaner.
“All right, so the comparison here was going from (pretty
highly regarded) Foobar 0.8.3 Secret Rabbit Code (best sinc,
slow mode) to playing the same files processed by Saracon
(input files are 32-bit float from the Sony pre-processing; output
files from Saracon are 32-bit, files are post-processed in Wavelab
to normalize to 0 dB and then dither to 24 bits using Wavelab's
Apogee UV-22HR. The Saracon-processed files were also played
back in Foobar, through the chain described above.
Note that one difference between the Secret Rabbit and
Saracon scenarios is that the (32-bit/44.1 kHz) files I was
playing back and upsampling in real time with Secret Rabbit
had a peak level at -6 dB, give or take, whereas the files
offline-processed by Saracon had then been normalized back
up to 0 dB. So the Secret Rabbit might have had a 1-bit
disadvantage at the point where the stream gets converted from
24 bits to 16 bits by the first Meridian 518 (the reason for
this, once again, is that EDR*S is strictly 16 bits in and 16 bits
out).
“OK, so what's the verdict?”
I’d guess whichever item better synergizes with the rest of the setup would be preferable. But it would be almost impossible to compare two conversions, if there is a ghastly number of conversions surrounding those being compared. The cumulative conversions as a whole have more effect on the sound than changing just one of them in the chain, provided the conversions being compared otherwise do the job properly.
If you’re a brave soul, you should try playing the same source with zero conversions. Process your original signal to 24/96 (if it happens to be analog), and play it back native. As I mentioned earlier. I’m certain you'll get a much purer signal than what you're currently doing. And the difference might stun you.
> A 44.1-> 48 conversion at best wouldn’t sound all
> that great………. This is exactly what Microsoft kmixer
> does with CDs. . .
For some value of "exactly". A kid's record player does
"exactly" what a $50,000 vinyl rig does -- for some value
of "exactly."
I've **heard** the 44.1-> 48 conversion through Saracon.
It does sound "all that great". I wasn't, as I said, expecting
so much from this initial conversion (I upsample again
from 48-> 96, and then finally from 96-> 192.)
> The “click and crackle removal” suggests you’re sampling
> the music off vinyl.
No, the "click and crackle removal" I'm using on ripped CDs.
Hopefully, 99% of the time (at the extremely conservative
setting I'm using) it's doing nothing. Very occasionally,
it's getting rid of a click that would play havoc with
renormalizations of peak levels later on.
I was once using more aggressive levels (the preset suggested
for vinyl) because of what I seemed to hear as a
"smoothing" effect on the sound. Piano, for instance, got
"creamier". Yes, it was probably rounding off transients,
and I stopped doing it. I don't miss that effect, with
the latest chain -- I get the "creaminess" elsewhere, and
keep the transients.
> This is insane……… There are so many conversions (I counted five),
> it makes me cringe…….
I **knew** you, of all people, would appreciate it! ;->
> You’re in essence taking a 32-bit signal, downconverting it a
> few times to 16/44.1, and then upsampling it a couple times back
> to 24/96.
No. I can tell you're not that familiar with how Digital Audio
Workstations operate.
The **original** source is a 16-bit/44.1kHz CD, ripped to a hard
drive (with Exact Audio Copy, if that matters to you).
A DAW such as Steinberg Wavelab (or Sony Soundforge, or Adobe Audition,
or Cakewalk Sonar -- take your pick!), when it does **anything**
to the original signal (say, just changing the gain by 1 dB)
outputs a 32-bit floating-point stream (or even a 64-bit floating-point
stream, in some cases) to the next plug-in in its Effects (FX)
chain. If you were going to take the signal right out and burn
a new CD, you'd have to convert that 32-bit floating-point stream
back to 16-bit fixed point (via the application of dither, either
flat or possibly noise-shaped; it's a matter of hotly-disputed
taste what style of dither to use, and which sounds best:
Apogee UV-22, POW-R, MBIT+, etc., etc. At least its hotly-
disputed when you're dithering to 16 bits; the mastering
engineers care less about the details when you're dithering
to 24 bits, but you **should** still dither rather than
simply truncating, it's the "correct" procedure.)
If you're doing repeated processing on a digital stream, it's best
to avoid redithering until the very last step, if at all
possible. So if you've got Effect A that inputs 16 bits and
outputs 32 bits, followed by Effect B that can input 32 bits;
then it's best to keep the output of Effect A at 32 bits
to send to the 32-bit input of Effect B.
In practice, what you do will depend on the capabilities (or lack
thereof) of the boxes in your chain.
My initial processing (declipping and the dreaded click and crackle
removal) is in WaveLab, which takes in the 16 bits (at 44.1 kHz)
and **can** output 32 bits. The next thing in the chain is
Saracon, which **can** input 32 bits. So I keep the output of
WaveLab at 32 bits. Get it?
When it gets to the soundcard to be played back (through the DSPs
which are hardware boxes, and only operate in real time,
such as EDR*S), then I have to dither the 32 bits down to 24-bit
fixed at that point. I do **not** "downsample" the sampling
rate! The sampling rate is always going up, up, up!
The most radical bit-depth reduction is required by that EDR*S
system, which unfortunately is 16 bits in and 16 bits out.
(I thought when I acquired it that it would be 24 bits in and
out; nobody, not even the erstwhile Audio Alchemy folks,
really remembered the details from more than a decade ago. But
it ain't -- it's 16 bits.) So, I'm using the Meridian 518
(a well-regarded processor!) to do the dithering of the
24 bits from the soundcard back down to 16 bits for the trip
through EDR*S. The sample rate has been increased to 48 kHz
at this point, so the noise-shaping algorithm has a bit
more bandwidth to work with (that's a good thing!).
The pseudo-expansion from 16 bits to 24 bits by the Meridian 518
at the other end is simply what audiophiles (as opposed
to studio engineers) were doing with 518's at home back in
the late 90's. It was discovered that if your DAC can take
more than 16 bits in, then just "filling in" the lower 4
(for a 20-bit DAC) or the lower 8 (for a 24-bit "DAC",
which is what I'm doing -- there weren't many 24-bit DACS
back then) makes the sound "better" (increases ambience --
it's a subtle effect). The Genesis Digital Lens (and
for that matter, the Audio Alchemy DTI Pro32) all used
this "trick".
The next step in the chain is the dCS Purcell upsampling
from 48 kHz to 96 kHz. I'll be trying out Saracon in place
of that, too. I may not be able to keep it up -- that
will require **two** dubs instead of just one.
> I guess if you like how it sounds, any technical criticism
> shouldn’t matter…… But then again, I'm afraid you might not
> realize what you’re hearing…………
Look, I'm not exactly a babe in the woods here. I've got
a ridiculous number of DACs in the closet. Dare I enumerate
them? MSB Platinum III. Audio Aero Prima. Lector Digicode 2.24.
Camelot Uther Mk. IV. Cary 306 SACD (and the Cary Cinema 11).
Tube Technology Fulcrum. Esoteric D-05.
Right now, **all** my systems are on an optical bus,
sourced from a computer. The master clock from an Apogee
Rosetta 200 is fed back to the sound card (an EMU 1212)
through 192 kHz ADAT optical. Each system "taps" off
that long Hosa ADAT-over-Toslink bus via its own
Apogee Big Ben.
> All you need to do is sample your original source signal
> (the very first step) at 24/96 (if it’s analog), and play
> it back at the native rate. No conversions necessary. The
> playback will be a *lot* cleaner.
OK, now we've seemingly strayed into the realm of digitizing
LPs (or recording live music). That's not what I'm talking
about here. I'm talking about upsampling (and processing
along the way, through some "magic boxes" -- Audio Alchemy
and Perpetual -- and some "magic software" --
Burwen Bobcat) 16-bit/44.1 CDs. You know, those familiar
silver coasters, that we around here never play anymore
except to pop 'em in the computer for a quick rip.
> If you’re a brave soul, you should try playing the same source
> with zero conversions.
Speaking of zero conversions, I also have three (count 'em!)
zero-oversampling DACs. A modest Scott Nixon, a pretty
modest Audio Note DAC-1 (kit), and a totally immodest
Audio Note Signature 4.1x.
I have not, however, listened to a "naked" CD at home
since 2005, when I first discovered the joys of this sort
of manipulation on the computer. At that time, I was
using Eximius DVD2One. It opened my eyes. Funny thing
is, though, when I posted my discovery here
http://db.audioasylum.com/cgi/m.mpl?forum=pcaudio&n=45351&highlight=DVD2One&r=
I got a lot of responses along the lines of (yeah, so?
We've been doing that sort of thing for **years**) ;->
http://db.audioasylum.com/cgi/m.mpl?forum=digital&n=111132
http://db.audioasylum.com/cgi/m.mpl?forum=digital&n=111172
And guess who posted a great big harrumph in response
to my article back then?
http://db.audioasylum.com/cgi/m.mpl?forum=digital&n=111131
http://db.audioasylum.com/cgi/m.mpl?forum=digital&n=111133
I don't understand the logic of spending $800+ for this element in the chain. The no of cables and connector mismatching (impedance) will influence the outcome considerably.
Personally, I would not want to spend so much time and effort logging and tracking the changes due to changes in one of the elements. I would also not downsample anything.
> I don't understand the logic of spending $800+ for this element
> in the chain.
But you see the logic in spending $7500 on a dCS Purcell (okay, maybe
$2500 used, these days). And I have a Purcell.
> The no of cables and connector mismatching (impedance) will influence
> the outcome considerably.
You may be right about that, at least in real time. Though the signal
is going through a **bunch** of reclockers (including, a fortiori,
the Apogee Big Ben -- not so much **because** it's a reclocker as because
it's the only thing I've found that will lock to the output
of the EDR*S boxes). The RME ADI-192DD (the first box after the
soundcard) is a reclocker -- again, not so much just to have a
reclocker, as because that box has the useful feature that if
its input drops out (as when I reboot the PC) it will continue
to provide S/PDIF at the current sample rate to the downstream
equipment, preventing me from having to spend half an hour getting
those ducks back in a row. Both Meridian 518s are reclockers,
as is the Perpetual P-1A.
And that Cary Cinema 11 tapping off the end for real-time monitoring
has sophisticated dejittering on its inputs, at least according
to their literature.
As far as cables and impedance mismatches are concerned -- you could
conceivably be right when the chain is listened to in real time.
But if you think that, when dubbing digitally from Computer A to
Computer B, if all the bits have arrived safely (as in **at all** --
without dropouts or glitches) they'll **still** "remember"
the impedance mismatches via which they got there when they're
freshly clocked off a hard disk drive in another playback system,
then I'm afraid we've parted company in our basic conceptions
of how this world works.
> I would also not downsample anything.
If you mean by "downsample" change the sampling rate from a higher
to a lower one, then I'm not doing that at all.
Bit-depth reduction is another story. That happens (hopefully
together with the proper application of dither) **any** time
you do DSP on a digital audio stream. Including in the digital
filter chips of CD players.
We shall differ.
I did not spend $7500 on my Purcell; it was $1500 used.
In retrospect, my 972 is a much better product and I wouldn't even spend $500 on a Purcell. Reason, it will not take more than 96k input; the 972 will.
...the weakest link in the chain is the speakers. cambridge soundworks and he's hearing a software resample from 44 to 48k through all that gobbledy gunk hardware? you've got to be kidding.
i'm afraid either the original poster is joking about all the front end hardware, or, Todd said it best, is 'insane'.
That Jim has lots of studio experience. Which is why he is a "bits is bits" believer and thinks he can process a digital signal with 10 different hardware boxes and have something BETTER come out the other end.
The fact he is altering the original signal and hearing differences is not questioned here. What is questionable is if the change all that hardware is imparting is a good change. Sounds to me like the word "creamier" implies "smoothed out" or "dummed down".
There is no mention of any concern for which box is clocking what, and what the sources for jitter are in that chain. The sheer number of digital cables involved (alone) and Toslink conversions makes one go "Hmmmm..."
I dunno. Sounds "neat" and everything but it leaves me wondering:
1) Is every process and SRC really needed?
2) Are so many boxes needed for the desired processes?
3) Is all the sample rate conversion required? Or just a necessary evil of getting so many different boxes to talk to eachother in the prescibed order?
4) Is Jim just "box happy" and trying to hook up everything in his collection at once to suggest that to play digital back properly you need a 4-foot tall road-rack full of digital processing gear?
Of course, we don't know what Jim is hearing, but how he is managing to get a low jitter signal out of that lalapalooza of gear is contray to my own experience anyways. I find less is more with digital transmission and conversion. But YMMV, of course.
You sure got some nice gear there Jim. The number of iterations for a digital front end you have there are limitless - and somewhat baffling! I'd never get any sleep if I had that much gear to play with.
I do agree with the other posters though. Compare your meag-stack to a "less is more" approach and see if you're really doing your "bits" a favor or not.
I think Todd is on the right track here, and I don't always agree with Todd!
Cheers,
Presto
> That Jim has lots of studio experience.
No studio experience whatsoever, in fact.
> Which is why he is a "bits is bits" believer. . .
I dunno, I would have thought that a "bits is bits" believer
would be inclined to think that what you get out of a mass-market
CD player is as good as you can (or ought to) expect.
> There is no mention of any concern for which box is clocking what,
> and what the sources for jitter are in that chain.
In that chain, the sound card in the computer is obviously
the master clock. Perhaps not optimum for "serious listening",
but for workstation use (getting the bits from Computer A to
Computer B through Box 1, Box 2, etc...) perfectly adequate
(I do not have loss-of-sync or dropout problems.)
Each box is clocked by its own S/PDIF (or I2S) input.
That's what S/PDIF does, you know -- it embeds the clock
in the data stream.
Nevertheless, I was able to hear what I heard, while monitoring
that "workstation" system.
When I play back the **end product**, my clocking is beyond
criticism. An Apogee Rosetta 200 feeds its clock **back**
to a sound card (EMU 1212), and serves as the master clock
(on an ADAT optical bus, reclocked for each system by that
system's own Apogee Big Ben). Is that concern enough?
> The sheer number of digital cables involved (alone) and Toslink
> conversions makes one go "Hmmmm..."
For workstation use, this is irrelevant, as long as none
of the data is actually lost. For real-time listening -- well,
some (most) of those boxes **are** reclockers.
You know, there was a real fad a while ago for "stacking"
reclockers. Two DTIs in a row? Wow! Six? Awesome! ;->
> and thinks he can process a digital signal with 10 different
> hardware boxes. . .
Yep. And these are not strictly studio boxes, either. They
were primarily marketed to audiophiles:
EDR*S -- a prototype of a system Audio Alchemy never got to
build. Based on the DTI Pro32, which was marketed to audiophiles.
In a more flexible form, this might have been of use to studios.
Perpetual P-1A -- primarily marketed to audiophiles, though
some studios have found it useful.
dCS Purcell -- created (as the dCS 972) as a studio tool,
until a genius (abetted by Jonathan Scull's review in Stereophile)
found a way to market it to audiophiles.
Meridian 518 -- this is (or was; it doesn't do more than 48 kHz)
a serious tool for CD mastering, but was also used by audiophiles
for "sweetening" the sound of CD playback. Can't really tell
what Meridian's marketing intentions were, though.
Hey, I don't expect anybody to take me particularly seriously,
here. I'm just having fun!
Thanks for the info Jim.
I confused your processing chain with your playback chain. Duh. sorry.
Yes, slaving a EMU1212 to a Apogee Rosetta 200 *does* indeed sound like a solid clocking scheme.
At first I thought you were running through all those processing boxes live...
Cheers,
Presto
> At first I thought you were running through
> all those processing boxes live...
Well, in the mini-review I posted of Saracon software SRC,
I **was** listening to all those processing boxes live.
(When I'm sittin' in front of the computer doin' the thing,
I'm monitoring that chain through a Cary Cinema 11 DAC
hooked up to a Cambridge SoundWorks computer speaker
system. Also used to watch TV. ;-> )
- 24/44.1 Toslink S/PDIF out from Prodigy 7.1, ASIO driver
- to an RME ADI-192DD
- thence to a Meridian 518 set to reduce 24-> 16 bits
- thence to the Audio Alchemy EDR*S chain
- thence to an Apogee Big Ben
- thence to second Meridian 518 set to expand 16-> 24 bits
- thence to a dCS Purcell doing SRC from 48 to 96
- thence to an Assemblage D2D-1 in "Transparent" mode
- thence to a Perpetual P-1A set to run its "Resolution Enhancement"
- finally to the Cary Cinema 11 at 24/96
Phew. That's quite a list. So help me out here: why go from 16 to 24 bits with software, then spit out 24 bits from the Prodigy only to go back down to 16 with Meridian 518 #1 and then back up to 24 again with Meridian #2? Methinks staying at 16 bit then using only the second Meridian in the chain would be a good place to start... a slightly more minimalistic approach?
And... why did the Big Ben go where you put it in the chain? I mean, why reclock there but then rely on device input PLLs at every stage downstream in the chain? Or are you using the big ben as a masterclock for downstream as well?
Thoughts?
Cheers,
Presto
> - 24/44.1 Toslink S/PDIF out from Prodigy 7.1, ASIO driver
This should have been written
"24/48 Toslink S/PDIF out from Prodigy 7.1, ASIO driver".
(At this point, I've already run Saracon on the 32/44.1 file from
Wavelab.)
> why go from 16 to 24 bits with software, then spit out 24 bits
> from the Prodigy. . .
I go down to 24 bits fixed-point from the 32-bit floating-point output of Saracon
because that's the widest output any sound card can take.
> . . .only to go back down to 16 with Meridian 518 #1
Because of the limitations of Audio Alchemy EDR*S. It's strictly
16 bits in and 16 bits out (I had to learn this the hard way --
nobody told me this beforehand.) I **really** like the sound
of EDR*S, and don't want to do without it in the chain.
See the Usenet post from 1996 by somebody who heard the EDR*S
system at the Stereophile show in New York in 1996:
http://groups.google.com/group/rec.audio.high-end/browse_thread/thread/44659620494e458e/90d03717d9b252bd
(Gabe Wiener and Steve Zipser are no longer with us.)
There was also a guy named Thomas W. Shea -- quite a serious record collector,
apparently -- who had an EDR*S system built for him by Audio Alchemy. I tried
to contact him about it, but he's disappeared. There were some letters
from him about all this posted in Stereophile, back at the time.
He also posted an inquiry on Usenet:
http://groups.google.com/group/rec.audio.pro/msg/02681dcf8123b7bf
(He never realized, poor guy, that he didn't **need** a 20 bit recording
capability to capture the EDR*S output, since it only outputs 16 bits. Of course,
he didn't have access to software DAWs with their bit meters, either.)
Meridian #1 dithers the 24 bits from the sound card back down to 16 bits,
noise-shaping over the bandwidth provided by the 48 kHz sample rate,
and then feeds those 16 bits to the first EDR*S box.
> And... why did the Big Ben go where you put it in the chain?
The Big Ben is where it is because the output of EDR*S is **extremely**
dirty. Peter Madnick warned me about this -- he said I'd probably
have to play musical chairs with the consituent 8 DTI Pro-32 boxes
until I happened to stumble on a sequencing that would produce
a usable output from the last DTI's S/PDIF output (the boxes themselves
are all hooked together with I2S; the first one also gets S/PDIF).
Even an RME ADI-192DD couldn't sync to EDR*S. Fortunately, the
Big Ben has no trouble syncing, and even achieves narrow lock.
Phew! And it worked the first time -- I didn't have to waste hours
or days playing Permutation City with the 8 DTIs.
> Or are you using the big ben as a masterclock for downstream as well?
The only master clock (in the workstation chain -- we're not talking
about the EMU-1212 slaved to the Rosetta 200, here) is the
clock in the Prodigy sound card itself. Everything else is clocked
from its input, whether S/PDIF Toslink, S/PDIF coax, AES-EBU balanced,
or I2S (and they're **all** there ;-> ).
Even with the Big Ben, though, the EDR*S boxes are **very**
touchy. Once they're up and running, they're stable for days.
But if the power goes off, it's a **major** Pain In The A**
to get them all going again -- some of them will not boot up
if they're connected to their neighbors. So its a matter
of starting them up one at a time, and then connecting them
up to their neighbors via the short I2S cables after each one's
DSP has booted successfully. If the boot is not successful,
all the LEDs on the box come on and the processor
just locks up. I had to discover all this more or less by
trial and error. This is **not** a commercial product!
I'm still amazed that Peter Madnick and Mark Schifter were
willing to do this for me -- I paid for Madnick's time and
use of the intellectual property, of course. He had to go
digging up the DSP code and then burn PROMs. I installed
the chips in the 8 DTIs myself, removing and saving the stock
ROM chips. One "little" glitch along the way -- in the
original PROM shipment Madnick sent me, **every other**
chip failed to work. There was some sort of systematic
problem with the partitioning of the code, apparently.
But he figured it out and sent me 4 "odd" ROMs (or "even"
ROMs, I can't remember which) that worked.
Yes, it was worth the trouble. ;->
"Go back down with Meridian #2" not #1.
If the EDR*S is close to your heart, that's cool - but why not keep your output at 16/44.1 (lose Meridian #1) and use Meridian #2 to get to 24 bit?
I just don't see the benefit to get 24/44.1 output just to go back down to 16/44.1 right away.
Anyways, it might be worth an a/b comparison...
Cheers,
Presto
> I just don't see the benefit to get 24/44.1 output just to
> go back down to 16/44.1 right away.
You should have said ". . .don't see the benefit to get
24/48 output just to go back down to 16/48 right away. . ."
> why not keep your output at 16/44.1 (lose Meridian #1). . .
I could stay at 44.1 and also do 16 bits out of the
sound card, of course. Or even continue to upsample to 48,
and still do 16 bits out of the sound card. Note however that I am
currently doing (and would be continuing to do) some processing even
if I dropped the sample-rate conversion to 48 kHz that I'm also currently
doing -- the initial declipping, etc. So one way or another
(either entirely on the computer or using both the computer
and the Meridian) I would still be dealing
with a digital audio stream whose bit-depth is
greater than 16 bits after that initial processing
(because that's what happens when you do
**any** arithmetic on 16-bit audio). So I'd still be
dithering it back down to 16 bits, **somewhere**.
It's just a trivial matter of where it happens. At the moment, I'm
taking 32-bit floating files (32 bits because of the arithmetic)
and dithering them to 24 bit fixed-point in Wavelab.
(I'm now archiving them at the 24-bit stage, too.) I could dither
them right down to 16 bits, though if I wanted to keep a
copy at 24 bits I'd then have **two** sets of files, and
an extra processing step. It's just more convenient to have
one set and do 32-> 24 on the computer, and then offload the
24-> 16 in real-time to the Meridian, since I have it
(the Meridian 518 was quite a respectable CD mastering
processor in its day.)
As far as the **sample rate** is concerned -- the upsampling
to 48 kHz ahead of EDR*S -- that actually
has to do with both the facts that:
1) one way or another, I **have** to bit-reduce when I'm
going into EDR*S if I'm going to be doing any pre-processing
on the ripped CDs.
I was majorly bummed out when I discovered that EDR*S
is only capable of 16 bits **out**. The stock DTI Pro 32
has a button you can press to select 16-, 20-, 22-, or 24-bit
output, or some such selection. But the buttons don't work
in EDR*S -- the program in ROM takes over completely, and
the DTI Pro32s become fixed-function processing blocks,
with no "user interface". My first clue to this, of course,
was that the buttons in the last box didn't do anything
(the EDR*S ROMs do go in a specific sequence in the chain).
Then I verified using Wavelab's bit-depth meter that it was
only, in fact, outputting 16 bits. And then I went to bed. ;->
I **assume** that the first EDR*S Pro32 is also only capable of
seeing 16-bit input, That **is** the safest assumption, because if I
gave it 24 bit data and it's only capable of seeing
16 bits, it would be **truncating** the lower 8 bits, and
that's much worse than properly dithering from 24 to 16.
Also, I read someplace that the stock DTI Pro32 could only
handle 16 bits in, except for a very late (possibly experimental)
version of the software. Possibly even post Audio Alchemy --
ex-employee Dusty Vawter was releasing revised DTI Pro32 ROMs,
and supporting DTI Pro32 modifications and repairs, even after
the company ceased to exist. Anyway, I somehow doubt
that EDR*S is using any of Dusty Vawter's special sauce.
The interconnections **between** the EDR*S boxes, via I2S,
are full 24-bit data paths, according to Peter Madnick.
Its just in and out that's limited to 16 bits. It makes
sense, because that show demonstration was all about "remastering"
CDs, not processing 20 or 24 bit recordings (which were
still few and far between in 1996; only the studios
had that kind of gear).
The next generation of the consumer version of that
"resolution enhancement" software, in the Perpetual P-1A,
**does** take 24 bits in and send 24 bits out.
In fact, it expects to be working at 24 bits and 96 kHz.
The Analog Devices SHARC 21065L DSP processor in the P-1A
is also more powerful than the older Texas Instruments
TMS320C31 in the DTI Pro32.
But Madnick et al. didn't really remember any
of this in detail, and I was not inclined to bother
Keith Allsop, the DSP programmer, who also may very
well not have remembered. **I** certainly don't
remember the fine details of any of the programs I
worked on 15 years ago!
Anyway, since it's necessary, bit reducing (and noise-shaping) over
a 48 kHz bandwidth does slightly less damage than doing
it over a 44.1 kHz bandwidth. The artifacts are just a
little bit further away from the audible band. And
2) the processing inside of EDR*S also has just a
little bit more "breathing room" at the 48 kHz sampling
rate. The system will do 48 kHz, so why not take
advantage of it?
(It doesn't matter so much that I'm doing the pre-processing
in Wavelab as the first step, at 44.1 kHz, precisely because
I'm **not** bit-reducing before sending the files on to
Saracon. I'm handing Saracon 32-bit files.)
> . . .that "resolution enhancement" software. . .
Speaking of which, there was some earlier discussion (from 2003)
about that here on the Asylum. With a picture. ;->
(The Audio Alchemy DTI Pro was an earlier version of the Pro 32.
It used the Star Semiconductor SPROC (fixed-point, quad processor) DSP,
but when that supplier went away, Audio Alchemy recoded the program
for the Texas Instruments 32-bit floating-point DSP).
http://db.audioasylum.com/cgi/m.mpl?forum=digital&n=57554
And a couple of years earlier, both Keith Allsop and Mark Schifter
participated in an Asylum thread about the P-1A:
Re: The Review in the current Stereophile...
Posted by Keith Allsop on January 9, 2001
http://db.audioasylum.com/cgi/m.mpl?forum=digital&n=15064
Re: dither vs. interpolation
Posted by mlschifter on January 5, 2001
http://db.audioasylum.com/cgi/m.mpl?forum=digital&n=14893
And, of course, this whole topic is guaranteed to make some
folks around here foam at the mouth.
Earlier this year, there was a thread about av123, the folks
who sell (or once sold) the Perpetual gear.
Should Be Renamed "Junk Science"......
Posted by Todd Krieger on May 03, 2009
http://www.audioasylum.com/audio/general/messages/54/544059.html
"[O]n [the av123 forum]. . . one of the threads happens to be
titled "Interpolation: The Real Deal?", which is a discussion of
"interpolation vs. upsampling", the technically-corrupt topic
originally brought up in a Perpetual Tech. "white paper"......
Comparing upsampling to interpolation is like comparing a car to
an automobile....... Upsampling or oversampling is simply the means
to apply the interpolation...... Interpolation cannot be done without
upsampling or oversampling...... The premise that upsampling and
interpolation are two different methods of digital audio playback
(filtering) is total BS......"
Perpetual's (and earlier, Audio Alchemy's) use of the term
"interpolation" to describe what their "resolution enhancement"
program does, was a choice of terminology that made some folks
see red, and caused no end of confusion. I think a better
word might have been "extrapolation", but I am not
Mark Schifter.
> ...the weakest link in the chain is the speakers. cambridge soundworks
> and he's hearing a software resample from 44 to 48k through all that
> gobbledy gunk hardware? you've got to be kidding.
Nope, not kidding. Yep, hearing a software resample from 44 to 48k
through all that "gobbledy gunk" hardware.
The Cambridge Soundworks speakers happen to be the ones on the computer I've
been sitting in front of all weekend. Let's see. Would my main
downstairs system with Quad ESL-63's suit your taste?
I knew, BTW, that somebody would say "you've got to be kidding"
about the speakers. The funny thing is, I've learned by **long** experience
that it's quite possible to 1) enjoy the music, and 2) hear subtleties,
through quite modest single-driver speakers. No, they won't
shake the walls in a cathedral-sized living room. But from a foot
away? (And yes, over the whine of the disk drive and the fans;
I have a diskless, fanless PC for "serious" listening, but it's
hardly a workstation.)
It's also true, from my experience, that big speakers, from halfway
across the room, tend to **gloss over** subtleties. For one thing,
in the real world, you can't (for human reasons) turn them up
loud enough even to hear all the music over a typical household
background, let alone to hear equipment subleties.
And yes, I also listen to my Quads "near field". Like giant
headphones.
> i'm afraid either the original poster is joking about all the
> front end hardware, or, Todd said it best, is 'insane'.
I expected that somebody (or a lot of bodies) would accuse me
of pulling everyone's leg or trolling, too. As far as my sanity
(or lack thereof) is concerned -- well, you may be right
about that!
and you're right about nearfield listening being more revealing. however, i would be pretty ready to attribute your 'major upgrade experience' to a placebo effect. not that i think dsp, etc, is meaningless or can't be heard.
i've never heard any differences using, say, SRC on foobar but then my computer setup is not first rate.
have fun!
Have you tried comparing a file upsampled with Izotope/Wave Editor? You can do 96 or 192 and hear this difference. A steal at $79 IMO.
Steve N.
Edits: 12/28/09
> Have you tried comparing a file upsampled with
> Izotope/Wave Editor? . . . A steal at $79 IMO
No, actually I haven't.
I decided to spring the $850 for Saracon partly on the
basis of the graphs at
http://src.infinitewave.ca/
(slow rolloff, you know. ;-> )
and partly because if I'm going to invest so much time
in offline upsampling (I've already invested a considerable
amount of time just on the same couple-dozen albums, but
at some point I want to declare the experimental phase over
and get on with building a substantial library of converted
CDs without worrying about which tool and/or procedure
might be better) I wanted the peace of mind that I expected
to have by going with Saracon.
Also, of course, if you don't already have a Mac,
Wave Editor is going to be more than $79. ;->
However, I do have a Hackintosh of sorts: a hacked
OS-X Leopard running (not terribly quickly) in a VMWare
virtual machine. Maybe, just for giggles, I'll spring
for Wave Editor and install it there (I'll have to use
MacDrive to exchange the audio files between the virtual
Mac and the PC).
No need to spring for the $79. Audio File Engineering offers a free 15 day trial.
One thing, please report your findings.
Is there a batch processing mode in Izotope that would allow "automated" processing of several hundred files?
Check out Audiofile Engineering's Sample manager, I believe it does batch processing, although I've never used it.
I do use Wave Editor, it's sibling. Both use iZotope. AE's products are world class at consumer prices.
clay
It turns out that Sony Sound Forge Pro 10 has **exactly**
what you'd want.
It has a Batch Converter that allows you to build a list of
files, and build a list of processes to run on each of them.
Looks pretty simple to use.
**And** it comes with iZotope 64-bit SRC! (I hadn't realized
that.) And also, of course, iZotope MBIT+ dither.
Both available from the Batch Converter menu.
Not **exactly** cheap at $400 (from Sweetwater; it might
be possible to get it for a few dollars less elsewhere),
but it will do exactly what you want.
> It turns out that Sony Sound Forge Pro 10 has **exactly**
> what you'd want. . . It has a Batch Converter. . .
> **And** it comes with iZotope 64-bit SRC!
Except that the iZotope SRC apparently doesn't work in Batch Converter,
according to Tony Lauck (see above):
"But note, as I said in another post, that iZotope SRC
and/or iZotope MBIT processing didn't run properly under
batch converter due to a bug in Soundforge (version 10.0a)."
Bummer.
There is a version of Wave Editor that runs batch mode.
Steve N.
Well, Weiss Saracon is a standalone program, and it **does**
do batch processing, and the batch processing works fine
(I've used it).
iZotope 64-bit SRC is not a stand-alone program.
The "iZotope RX Advanced" (stand-alone) audio restoration
program ($1200, for Windows or Mac OS x) **includes**
iZotope 64-bit SRC as a function of the program, so
whether the SRC could be used in batch mode would depend
on the batch capabilities of the host program. iZotope's
blurb sheet says "Batch processing makes applying
multiple processes to large files much less time consuming."
so it has **something** they're calling "batch processing",
but it's hard to know exactly what it will do without
trying it out.
There's also a cheaper "iZotope RX" which has an SRC,
but it's not the super-accurate "64-bit SRC".
I assume you're really asking about Wave Editor, the $79
DAW that includes iZotope 64-bit SRC (Mac only).
Well, again, the batch-processing capabilities of the SRC
would depend on the batch-processing capabilities of
the host DAW, and the documentation at
http://www.audiofile-engineering.com/support/manuals/we/1/html/index.html
does not look too promising.
There are other ("lesser") stand-alone SRC programs that
do batch processing. Like r8brain.
> Baseline SRC to 48 kHz using Secret Rabbit Code,
> "best sinc", slow-mode on; final Foobar output dithered
> to 24 bits with "Strong ATH noise shaping". 24/44.1 Toslink
> S/PDIF out from sound card. . .
This should read "24/48 Toslink S/PDIF out from sound card. . ."
of course.
> I've been listening through the "front end" of my signal-processing
> chain, in real time. . . [finally] to a Perpetual P-1A set to run its
> "Resolution Enhancement" program at 24/96, and [then] to the Cary
> [D/A converter].
I should also note that the P-1A reduces everything going through
it by 6 dB, so in all cases (even files going in at 0 dB peak)
the data getting to the D/A converter's own digital filter is
at -6 dB or less.
FWIW.
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 ]
