|
.
|
Myth:
Digital sound doesn't
lose quality when transferred.
|
"We throw away thousands of
[blank CDRs] if they're not right. We get samples from manufacturers
and A/B. We listen very carefully before we pick the lot
that we're going to buy. A lot of people think once you're going
D to
D, it doesn't matter because it's all numbers. But you can hear it.
Every step makes a difference, and when you add all the subtleties up,
the result is dramatic." -"Big Bass"
Brian Gardner - Bernie Grundman Mastering
|
This is a controversial subject, but if you want your masters
to be the best, it's worth looking into - vs. just assuming any one
theory. There has been a long-held thought that digital can be
transferred, cloned, copied, converted etc. and the sound stays the
same. Numbers are numbers, right? They are... if no jitter, errors, or recalculations are
introduced when the data is transferred.
Granted, digital does what it does well because it sure is
convenient and it's put recording studios into the hands of a lot of
people. It just has a different set of limitations that we weren't
expecting when the technology came out. This is the same thing that
occurred when we discovered that solid-state electronics didn't have
some of the cool sonic qualities that tubes did.
First, let's start with some basic, simple definitions:
1) Jitter. What the heck is jitter?
Jitter is error in the time between digital samples. An
audio CD plays 44,100 samples (analysis snapshots) per second - but the
clock determines how evenly spaced together the samples are. The more
accurate the spacing from sample-to-sample, the better the sound. If
you were to record some music where the
original clock sampled IN evenly |--|--|--|--|--|--|--|--| but the
playback OUT was uneven (jittery) |---|-|--|-|--|---|-|--| the playback
will sound more brittle, less wide and smooth, more dry and
grainy. Lots of jitter can cause clicks or tics in the
sound.
One common place where jitter is heard is on some less-expensive
digital workstations or computer audio cards. This is why you
often hear about "reclocking" a system for the purposes of improving
the sound of a digital system. Computer-based digital audio cards
can have more jitter due to
ground and power contamination and the proximity of other clocks on the
computer's motherboard.
We've looked at clocking issues and transfer-to-glass speed
at the pressing plant stage too (although today's yellow-book DDP
masters
give us new results, particularly when pressing plants are use network
systems).
2) Errors - Audio CDRs can have
errors - and the vintage of the burner, the condition of the media all
can play a part of errors. When a DDP master is not being used,
pressing plants
(should)
always do BLER (Block Error) scans to make sure that uncorrectable
errors don't cause your CDR master to be rejected. Never mind fatal
errors caused by the laser turning on and off -- differences in media
can make a difference too - and different burners have different
performance characteristics.
3) Recalculations occur when you are
processing sound in any digital processor, like Pro Tools, Nuendo, a
Finalizer, plug ins, you name it. When we change a volume or eq control
on a Digital Audio Workstation, we think in analog terms - namely the
volume goes up or down, the top end gets brighter. In the true
analog world it's the same waveform being acted upon by that fader, eq
knob, etc. In the digital world, those changes cause changes in the
numbers - if you do a ton of processing to a signal, there's a lot of
recalculation going on and it can compromise the coherency of the
original waveform, causing it to loose solidity, stereo depth-of-field
or become thin and loose "life." New 64-bit systems may do a lot
to resolve aspects of recalculations.
4) Physical/fragmentation/transfer issues
- dropouts can be a common problem with some DAT tapes, but more common
is when files get fragmented (mostly slows access, may
cause crashes)... files become corrupted (software can be largely to
blame), cheap digital cable can have reflections and other
sound-altering problems... fingerprints and scratches on CDRs can cause
drop outs... fan vibration can rock a hard drive.... high-end digital
gear folks say that Toslink/fiber optic cable doesn't sound as good as
AES/EBU or BNC... Error correction helps, but when it comes to your
master, you want the ultimate... not second-best.
Different CDRs sound different. Even
within the same brand. Their error
rates vary, even with the kind of CD burner used. I've listened to
Kodak's gold "Writeable" CDRs, their gold "Audio" CDRs, their gold
"Recordable" CDRs, and each sounds different.
Even Maxell 700 mb
silver CDs sound different from the Maxell
Pro or Maxell "Music" CDRs (and those black ones sound harsh!!!). TDK's
sound different from Memorex, Mitsui, Verbatim, Sony... If you're
cutting CDs on a home computer, don't use the black ink-on-the-bottom
CDRs or generally cheap CDRs. You get what you pay for. Different
burners can have different laser power and the bottom of the CDR may
need something different to perform well. (If this concept pushes your
computer buttons, be sure to read this whole article.)
|
My favorites are Maxell 700 mb Music CDRS (80 minute gold) - for a
solid and balanced high-to-mid-to-bottom and wide sparkle • Maxell 80
Minute Pro (blue) for robust low end, detail and clean immediacy • Fuji
80 Minute Audio for a wetter sound (smoothes out the edges). • Memorex
Music 80 minutes is very nice • Taiyo Yuden 700 MB are close, the
Mitsui and BASF are in there, Apogees has a nice width but lacks
warmth, Experiment and see what you prefer!
|
Look this is all great, but I've got to interject something
new, and then we'll get back to the more tweaky stuff...
Different digital formats
sound different, and
...different hard drive systems sound different.
|
|
Creating Separations as your delivery method into mastering makes a huge
difference in the sound. Assign your tracks to subgroups, and
record them each separately in addition to your final mixdown. The Separations are re-combined in mastering
and here's a simple example:
|
|
The session: Semi-live jazz recorded in a really nice
studio via Nuendo vs.3 - 48k 24 bit, MOTU clock on record, Ryan Hafer,
engineer Bass, drums, acoustic piano, sax, vocal, maybe 16 tracks. New
PC computer with UA plug in cards, low (almost no) amount of CPU
"strain" - brought his computer up to the mastering room reclocked via
PSX-100 AES/EBU out to the mastering rig.
We tested the same song sent to the mastering console via these sources:
1. Direct live mix output from his tracks played out
of Nuendo AES/EBU digital output
2. 5 Separations (bass, drums, acoustic piano, sax,
vocal) created via loop-back S/PDIF (not internal bounce) onto internal
IDE drive -- recombined in new Nuendo mix session, same output
structure as #1
3. IDE separation files copied from the IDE drive to
a Firewire drive played out of a new Nuendo session in the same fashion
4. Firewire drive plugged into Digital Performer 4
mix session on Mac OS X 10.3.9
We loaded these into the Sonic (destination) mastering computer through
minor mastering processing, as it was a very good recording. Nothing
was touched as we loaded in each source version, and we heard the
following played back out of the Sonic computer (Prism ADA-8 D-A's) on
the same Edit Decision List:
|
|
1. Live mix sounded very good, but struck me as
being a hair cloudy.
2. Separations re-combined from internal IDE files
were awesome! Much much more air around the voice, more apparent level,
more width, better articulation of the piano notes, bass was more
audible and tighter, less fog around 300Hz, drum transients were
sweeter, more open, easier to hear the sound stage of the room the
drums were in, more front-to-back depth. The live mix sounded muted
compared with this when instantly A-B'd.
3. IDE Separations transferred to the Firewire drive
played out of Nuendo lost MUCH of the air and openness, but still
sounded better than the live mix. The image was shrunken compared to #2.
4. Firewire drive plugged into DP sounded almost
undetectably different from the Firewire on Nuendo, maybe more analog
sounding on bottom, slightly rounded transients, pleasant, but not an
improvement or deprovement... did I say that?
5. Firewire drive in DP eq'd via plug ins - drums
were more grainy but overall fidelity and "mastered" sound punched out
due to more aggressive eq changes. We didn't use this version, because
#2 had so much more air and spatiality to it. We performed some nice eq
on the Separations in Nuendo and the client was super excited about the
sonic results.
|
"I've done 20 albums in
my career, with major labels and smaller independent ones, and the
'Separation Mastering' process has by far provided the most superior
sonic results I've heard. I don't think I could have my music mastered
any other way now that I've experienced the extraordinary sounds
achieved through the Separation process in mastering music."
-Anthony DaLuz, San Diego, California |
"Mastering
with Separations sonically took our music to a level I wasn't
even aware it could get to. Now that we've used mastering with
Separations, we won't settle for anything else."
-Harlis
Sweetwater
|
Now for more controversy:
Burn speeds.
Years ago it was very apparent that the higher the burn speed, the
worse (and more unreliable) a CDR would sound. I still hear more people
complain, "My 12x CDR copy won't play on some systems and it sounds
harsh." We make 4x reference CDRs here when a fast copy is requested,
and golden-ears engineers can hear the difference.
|
"I thought everything was
perfect
on our project except the top was a hair edgy and the bottom needed a
touch more support. When you recommended that I check the master
instead of the copy, I was surprised - the tone I wanted was exactly
right on the master!"
-Matt Forger -
renowned veteran engineer
|
I've experienced that burning CDRs at 2X sounds
different than 1X. I invited a professional engineer and a
stereophile guy to listen to the same album on two different CDRs...
one cut at 1X one at 2X. The engineer preferred the 1X, and thought the
CDRs had different mixes on them. The stereophile guy
simply felt the sound on the 1X was sweeter and wider. Burning CDRs at
higher speed (like 4X, 8X, etc.) adds hardness and sterility to the
highs and mid-highs. If this Frye's your circuits, click on the links
that follow later in this article...
Different findings: I
received an email that relayed info from Alesis:
"While this [burn speed issue] was true a few years ago, CD writers
have made leaps and bounds. Most produce less jitter at 4xs speed than
at 1xs. A large group of mastering engineers and critical listeners
agree that CDs cut in different ways tend to sound different. The CD
differs from other storage media in many ways, but the critical point
is that the timing of the output clock and the speed of the spinning
disc are related. The output of the CD player is a clocked interface,
and the data are clocked off the CD disc in a 'linear' fashion, one
block of data after another. A buffer is used, which theoretically
cleans up the timing to make it regular again, and, for the most part,
it does."
"Actually cdrmedia.com and ixbt-hardware.com acquired some real
expensive CD media testing equipment and did some of the most
exhaustive tests including the quintessential BLER tests in the
industry. Their conclusion was surprising and breaks the norm of
recording at lower speeds. According to their tests, today's high
recording speed burners tend to produce lower errors when the recording
is done at near their maximum rated speed or at least only a few
notches down."
My suggestion: Use your ears
- preferably when listening on a HIGH-RESOLUTION monitor system. I
don't think NS-10's will work for everyone in critical listening
situations. If your master mix is going onto a CDR, burn it at 1X (or
the slowest speed available). Enough cases have caused me to suggest
that the slower the burn speed, the less jitter. Make sure your
pressing plant gives you a reference CD so you can hear how accurately
their glass mastering translates to your final product. People have
different experiences - and it's not worth getting burned over! You
decide - or check with someone you trust and take their findings to
heart.
On my CDR masters, I make sure that I include the instructions, "Cut
glass at 1X ONLY." Even though some pressing plants will say that
cutting a glass master at 2X creates fewer errors (and saves them
valuable glass-mastering time), I insist on 1X, just as all the
major mastering guys do. Check out Roger Nichols' article " CDs give me the
Jitters!" where he gives you the nitty-gritty about the Steely Dan
album glass master experience.
Bob
Katz stated, "We believe that irregular pit depth or
irregularity of spacing of the pits (due to the quality of clocking at
different speeds) causes the servo mechanism in the CD player to work
harder and pass this onto the power supply for the D/A converter (in a
cheap player)." Remember - digital technology changes fast - keep tabs
on the available info.
One studio owner asked me why a CD copied to his hard drive and then
just burnt to CDR sounded so different. Aside from burning speed
issues, I got some input from a computer programmer that could shed
some light: Software designers are sometimes required to program for
efficiency instead of precision. This means in order to make a program
run fast at a competitive price, they end up making it handle larger
blocks of data at once, compromising the possible resolution of the
sound. A small error within a block gets corrected, and thus the whole
block is changed.
Different DAWs sound different. A
great engineer, Lynn Fuston, took a top-of-the line multitrack digital
studio recording and made 30, count them 30, identical (plus-or-minus a
couple microscopic hairs) mixes on different DAWs. This is the Awesome
DAWSUM comparison. Here's the 3D
forum home where you can check it out for yourself. Lynn also
has great "shoot-outs" on mic pre's and more.
DAT safety "clones" don't sound the same. Unless you are using a
high-end digital cable ($350 to $700) that eliminates cable
reflections, or you are synching your machines to a word clock unit,
DAT safety "clones" will sound more brittle in the highs, and have less
depth in the reverbs and room sounds. • All ADAT owners I know don't
like the sound of their backups as much as the originals, which can be
the pits if you're getting dropouts on your originals due to tape wear.
I've experienced transferring a sound file from one hard drive system
to
another and the sound changed. There are those who say that's not
possible, but here's some examples on how this happened:
(1) A few years ago, I copied a
Sonic Solutions sound file from one drive to another using the "copy"
utility in Sonic. I put the copied file into an edit decisions list,
frankly expecting it to sound exactly the same. When I listened, I was
shocked to hear it sound more sterile and harsh sounding. Thinking I
was nuts, I put the original file next to the copy and compared.
The sound had definitely changed. I was bummed because I needed to
clear off a hard drive, and I was hoping I could move the files and do
a low level format of the cleared drive. Could it be that the Sonic's
Copy Utility induced jitter into the sound file, and the difference had
nothing to do with the drive? Yep - could be...but wait... is it
possible that different containment, location and physical access
characteristics make a difference in the sound, even if the file is an
exact clone? If the two drives play back the data exactly the same, it
could sound the same... but what if one drive is a bit funky? Some say
that the file goes from hard drive to buffer before getting to the i/o
card, but could there be other factors?
(2) Again, a few years back, I spoke
with someone who knew of a major LA film editor who only bought certain
hard drives because he could hear the difference in brands. Fact or
fiction? Could differences in hard drive shielding or the proximity of
power supplies to cables have any effect? Could his monitor system be
so revealing that he could in fact hear these differences?
(3) Recently I did 3 projects where
I loaded a client's AIFF files into my Mac DAW. The files sounded fine,
but I saved them to my hard drive as backup. I then loaded the same
files from the hard drive so that I had a file from there as well as
from the CDR on my screen. I listened to both versions, and the hard
drive copy sounded better than the original on the CDR. Whoa! There's a
turnaround for ya.... the copy sounded better, not worse.
Nothing changed - same computer, same converters, same mastering signal
path - just two versions of the same file from different sources.
Granted, the difference in sound wouldn't keep you from selling any
copies.... Could it be that the CD-ROM drive was less stable than the
hard drive? Could the smoothness of data transfer affect the jitter?
(4) Even more recently, I had a
client with a Firewire drive with 24 bit AIFF files. He transferred a
song from one drive to another. Again, we loaded both the Firewire file
and the Mac hard drive file onto my DAW so each could be instantly
selected. With morbid curiosity, I stood in front of my console where I
could not see which file he played - and in 3 out of 4 comparisons, I
heard a difference. The Firewire file was warmer in the vocal and the
snare was less dry sounding.
Now trust me, the difference was very very very slight. It was
not enough to even remotely consider not backing up your files and
using them if you needed to! In case #(1)
above, the difference in sound quality was way worse. If case #(1) occurred again today, I would not
hesitate to transfer the file with today's modern hard drives
(particularly after I had listened to the copies just to be sure....)
The difference here in case #(4) isn't
anywhere close to the difference that occurs in other situations I talk
about on this page, particularly with DATs and digital cables, faster
burning speeds etc.
But let me just re-mention the part about backing up files so we're all
clear about this - ALWAYS back up your files. Do not ever use this
article as a reason to risk your projects data. Things can and do
happen, so always protect your time and money by backing up your discs,
your songs, your files, your bank statements, your pets... you can't
back up enough!
Professionals I've communicated with on
this issue have graciously shared their findings and I appreciate what
they've shared. Some state that changes and errors in data can
occur in many instances, but not for WAV files... I absolutely respect
what they have experienced, and it's ok if they show different
findings. It's the dialog that counts so that we're all a
little less assumptive and we keep looking at ways to get
better sound.
This is a controversial issue, and yours truly has even gotten a few flaming emails over this one! My purpose here
is to share what I've heard and experienced - I'm not here to
have an "I'm right" contest. Sometimes we can get caught up in
"ultimate" right and wrong, good and bad - and if we're getting all
upset over it, we're forgetting that this stuff is supposed to be about
music and technology and it's supposed to be enjoyable! My view is that
it's about what works and what doesn't work
for you that counts.
What about physical vibration from sound, fans, power
supplies....
[20 year computer engineer joins the
discussion] "After hearing this story and deciding not to chalk it up
to BS, I had some very clear possibilities -
#1 The firewire drive, was it external? Well, right there you have a
big difference, the power feeding that drive, and it's electrical noise
is separated from the computer.
#2 The internal drive, being extremely close to rest of things in the
machine, like the soundcard. (Even if it's like a rme) Can cause w/c
problems. w/c doesn't have error correction, it's grounded in the
machine with everything else. If I "wiggle" a cheap w/c cable and look
at the signal on my scope, I see the signal get nasty. In PC clones,
those 50$ power supplies are horrible. This is why you just don't see
people doing a/d/a in a sound card with much success. My belief (and
granted this is a guess so please feel to prove me wrong), is that the
w/c has the possibility of being effected because of the kind of signal
it is. This also applied to the "take it off the floor statement".
Aside from the obvious vibration differences in sound waves and fan's
etc. Imagine if the w/c cable wasn't moving so much because of
difference levels of vibration.. "
Meanwhile: The Pro Tools studio next
door did an experiment with me. We took a digital cable directly from
his rig and recorded a song into Sonic Solutions. Next we took the same
mix and loaded it into my system off a CDR cut directly on his
computer. We then compared the sound of the two files on my system.
AMAZING. The exact mix brought in via the direct cable version sounded
fuller, wider, smoother, more open, more detailed, more musical and
expressive. The CDR version sounded choked, grainy, and had less width.
It wasn't even subtle, and all three of the engineers in the room
agreed.
Reason: When digital audio
workstations (DAW's) render a stereo file of your mix, the
process requires a mix engine in the system to create that
stereo file. This is different from the mix buss which is
simply your stereo output. I have yet to hear a mix engine on any
system that doesn't degrade the sound somewhat. Solution: Go digital out from your mixdown mode
into a 24 bit Masterlink or 24 bit Tascam DAT machine. For mastering
you may want to bring in your DAW as your stereo source master (more here). Removing the master stereo fader
(so long as it's set to unity gain and you're not getting any
clipping/overs) can improve the sound too.
Q) Pro Tools HD now has a convert after
bounce feature to eliminate automation errors and a 48 bit mixer
plug in on the master buss. What's your opinion on using this to make a
stereo AIFF or WAV file for mastering? -Jim
A 48 bit mix engine would be very desirable. I would have to hear an
AIFF file made in that fashion up next to the actual output of the
"live" stereo mixdown to tell you if the comparison is better than
before. Typically when the stereo mix is internally made into a stereo
file, it loses some fidelity. If your choices are 24 bit DAT from the
stereo mix vs. 24 bit stereo AIFF file, I'd say send both and let's
here which sounds better. If only a 16 bit DAT machine or CD burner was
available, anything 24 bit is better. However if the AIFF was 16 bit,
again I'd want to hear both to see which is better.
If you are recording from a digital source like a computer or a hard
drive system like the Roland 1680 (Akai's and Tascam's are my favorite)
onto a DAT or stand-alone CD burner, be sure to use the best digital
cables you can afford. Cables make a significant difference - a $350
S/PDIF cable makes as much difference as a $3,000 converter.
Set up: We are "imprinted"
with the experience of analog music media. In other words, we are
accustomed to an analog signal remaining the same when an analog volume
control is moved. We are used to analog equalizers that shape the
waveform vs. digital processors which internally recalculate the
numeric data which creates an entirely new set of numbers (word length
changes, etc.). Tweak the daylights out of a track and you can lose
some of the original coherence to the sound. But remember, there are no
rules here, so you may want that exact result.
If you're a musician/studio owner, remember that none of this technical
advice may mean as much as the heart and soul of your music -
your musicianship - the actual thing that all of this stuff is designed
to reproduce and deliver to the audience who wants to hear you. When
you're considering upgrading to the next piece of gear... consider if
additional practice will improve your sound even more than the
slick new feature-filled Platinum Pro Version 39 Giga-Gizmo!
Now here's some technical computer-type debate
on this subject (and then some more music-based suggestions follow....)
Q) Transferring data from one drive to
another can change the sound? Where on earth did you learn that? You
know that a copy of files on a hard drive makes a 100% identical copy!
How can the sound change?
First of all, I appreciate it when someone questions me when they feel
differently. Different experiences make this planet interesting, and I
think we're all on the same page about wanting the best possible
quality. It helps to have a great monitor system that reveals subtle
details, and I have 35 years of music industry background as part of my
ability to listen critically. Some esteemed engineers say their isn't a
difference when a file is on a different drive - so how can the same
file on the same drive in the same computer monitored
via the same converters sound different
when the computer is physically supported differently?
Two other engineers (in session with me)
heard the sound change when we raised the client's computer off the
floor with soft isolation pads. The only thing that changed was what
the computer was sitting on. We found that setting the computer on a
hardwood floor made the sound more immediate and crisp, compared with
setting it on soft isolators. On the floor, the snare sounded punchier,
the kick more immediate, and the overall sound was tighter. A solid
platform is even better when vibration isolators are used - and you'll
be amazed at the difference a great power cable makes too.
Many audiophile stores sell suspension stands that use the principle of
supporting gear on a solid foundation, saying that everything from
power amps to CD players sound better if they're not on soft
material. Fact or fiction? Speaker cables sound better when they're not
lying on artificial carpet or plastic tile. Fact of fiction? Try it out
for yourself on a great monitor system and you decide. This page is an
invitation to be open to more than what's usually expected from digital!
CD audio vs. CD data storage
(quoted from an email from Lee du-Caine):
The reason a data CDR doesn't have errors and an audio CDR does is
partly because when the data CDR encounters an error, it re-reads the
section over and over until it reads the data correctly, whereas when
playing an audio CDR, this doesn't happen because it cant keep stalling
the audio output so it can reread the data over and over!
Songs would keep repeating tiny sections every time they hit an error.
Some high-end audio CDR players can do this which use industrial CDROM
drives, they spin the CD a lot faster and use a large memory buffer to
keep the music going while the mechanism tries to re-read the problem
sections.
When a regular CD player encounters an error, it discards the dodgy
sample, and uses interpolation to "estimate" what it should have been.
Also a data CDR has much more effective error correction mechanisms
built into the data, since it would be of no use if data got corrupted
with every CDR! So when CD data format was invented, they came up with
some new ways to store the data which made errors easier to
detect/correct. Read more here on how data CDs
are stored.
Q) My PC can produce a bit-for-bit copy of
any CD (in good condition), including audio-CD, in 5 minutes. Less, if
the copy is just on hard disk, rather than on CDR. The error-free state
of this can be verified. If this is the case, how can an audiophile
claim that a digital copy degrades the sound quality, that a green
protective sleeve over the CD, or a stroke on the edge with a 'magic'
red pen can reduce errors to the point of audible increase of sound
quality? How can anyone claim that certain CDRs 'sound better' than
others?
Dramatically better in some cases and it's all about smoothness of data
transfer (better transfer=less jitter). I've listened to many many many
CDs. Every mastering engineer I know is FUSSY about what brand of CDRs
we use, because.... they sound different. Different CDRs in different
CD burners make different errors and there's more here: http://www.digido.com/chart.html
Then be sure to click on Bob Katz'
excellent article on Jitter. Here's an excerpt:
"When I started in this business, I was skeptical that there could be
sonic differences between CDs that demonstrably contained the same
data. But over time, I have learned to hear the subtle (but important)
sonic differences between jittery (and less jittery) CDs. Clients were
... complaining about sonic differences that by traditional scientific
theory should not exist. But the closer you look at the phenomenon of
jitter, the more you realize that even minute amounts of jitter are
audible, even through the FIFO (First in, First Out) buffer built into
every CD player."
" CDRs recorded on different types of machines sound different to my
ears. An AES/EBU (stand-alone) CD recorder produces inferior-sounding
CDs compared to a SCSI-based (computer) CD recorder. "
"I've also observed that a 4X-speed SCSI-based CDR copy sounds inferior
to a double-speed copy and yet again inferior to a 1X speed copy."
There's
so much stuff on the Digital Myth, I've split the article up into 2
sections - click here to see The Digital Myth
Part 2.
Must we analyze digital sound
under the microscope when a 13-year old won't know the difference
anyway?
|
