PT Mixing Vs Analog Mixing
LilYellowMonstr
tracks) out to an analog board either individually or in submixes... What kind
of board are you using? What sonic benefits do can you describe using the
board and its analog summing buss over mixing in the computer?
At what stage of console does it become worth it? I doubt that sending PT
tracks post plugins to a Mackie 32 8 with an SSL compressor on the main inserts
would do much justice, ey? What do you think would be a good medium priced
console that would surely put internal digital summing to shame? thanks
John
Ethan Winer
> What do you think would be a good medium priced console that
would surely put internal digital summing to shame? <
I don't use ProTools so I can't speak to the quality of its
hardware. But I can tell you there's no inherent reason digital
summing should sound "worse" than summing in an analog console.
In fact, I'll argue digital should be better because digital
summing is simply adding a series of numbers. With analog
summing you have input and output impedances and capacitance and
power supply stiffness ad nauseum to contend with.
--Ethan
Monte P McGuire
Sometimes people use the inadequacies of an analog console to "gel" a
mix. This does not happen with digital, so if you need that, the
accuracies of digital are a liability.
Then again, one could add something like an Analog Channel or a DUY
shape plugin to the master bus and get loads of glue on a DAW...
Regards,
Monte McGuire
mcg...@world.std.com
Paul Winkler
Uh-oh, here we go again...
I'm not going to say what digital systems have "good" or "bad" summing
busses, or even claim that some sound better than others. I do not
have experience on which to base such an opinion. There have been
many, many arguments on this newsgroup about which systems sound good
or bad or whether the difference is even audible.
BUT as a sometime computer programmer I can tell you that it's simply
not the case that all addition is equivalent.
Many DSP systems use pure integer arithmetic, where any fractional
result is simply discarded. In other words, 1 / 2 = 0. In practice,
this works out more-or-less OK because we use a lot of bits and the
errors tend to accumulate in the lowest few bits. The advantage is
that, if we don't do any processing, the output can be guaranteed to
be identical to the input. There are two disadvantages: first, as you
do more processing, you accumulate more and more error and at some
point it will eventually be audible as distortion. Secondly, it's not
hard to overflow the limit on the largest number that can be
represented, causing clipping; third, the system gets less accurate at
lower volume levels, so if you have a large loss of gain followed by a
large make-up gain, you can lose a lot of accuracy very quickly. So
gain staging is important.
Many other systems (including most, if not all, "native" software
solutions running on a computer) use floating-point arithmetic. In
this case, 1.0 / 2.0 = 0.5, or (this is important) something PRETTY
CLOSE to 0.5, like maybe 0.49999999999999999. The disadvantage here
is that there's a slight inaccuracy right off the bat - many integers
simply cannot be perfectly represented as a floating point number. So
you might not be able to guarantee that the output is identical to the
input. But there are two advantages: one, the error tends to stay
fairly small even with extensive signal processing; and two, there is
a drastically larger practical range of numbers that can be
represented, so it's MUCH harder to cause clipping, and large gain
changes are not so destructive.
Furthermore, not all floating point processing is the same. A PC has a
choice between 32-bit floating point numbers and 64-bit "double
precision" floats. The 64-bit variety, obviously, is a lot more
accurate, but needs twice as much memory.
So there is a theoretical basis for arguing that different summing
busses do not sound the same. I'll leave it to others to argue over
which are "good" and which are "bad".
--Paul Winkler
LilYellowMonstr
"yeah I mixed in Pro Tools for about a year and then I tried mixing all
channels out the D/A out to an analog board and then threw that to some good
converters to a DAT and the difference between the analog mix and the digital
mix is night and day".
Digital mixing seems to be too complex, in that the summing buss issue is
really frustrating. For instance, I sat in with a guy yesterday who was mixing
nuendo tracks. He routes right out his 9652 card into a little Spirit Digital
board. The two mix is then recorded to two seperate tracks on the computer. I
asked him, why don't you just mix entirely within Nuendo, then he did this,
and we compared the mixes and the one summed through the 328 was undeniably
better, more space, more 3D. Yikes. Its just too much. I wonder if anyone
has tried doing this with Pro Tools and another DIGITAL mixer.
Mark McQuilken
(snip)
> In fact, I'll argue digital should be better because digital
> summing is simply adding a series of numbers.
(snip)
>
> --Ethan
Ethan:
I couldn't disagree more. It's not purely the act of adding--either in
analog or digital domains--that determines the "purity" of results. It's
the artifacts before, during and after the process that will determine
the sound. There are right ways and wrong ways to add signals together
in either domain. As always, it's the care exercised by the designer
that will determine the sound...
McQ
Ethan Winer
> Sometimes people use the inadequacies of an analog console to
"gel" a mix. This does not happen with digital, so if you need
that, the accuracies of digital are a liability. <
Now "inadequacies" is a concept I can surely agree with <smile>.
And I happen to agree with you about the "gel" part too. I've
recorded digital mixes to a cassette deck and liked the result
better than the original. But I never thought the cassette copy
was more accurate!
--Ethan
Ethan Winer
> BUT as a sometime computer programmer I can tell you that it's
simply not the case that all addition is equivalent. <
I've been a professional programmer for 20 years, and I don't
know of very many different ways to add 2 plus 2. I agree
integer math can have rounding errors during many computations.
But I was talking about summing, not signal processing. Using
8-byte double-precision values (64 bits) gets you VERY close to
ideal for the relatively modest precison digital audio requires.
> In this case, 1.0 / 2.0 = 0.5, or (this is important)
something PRETTY CLOSE to 0.5, like maybe 0.49999999999999999.
The disadvantage here is that there's a slight inaccuracy right
off the bat - many integers simply cannot be perfectly
represented as a floating point number. <
I really don't see why that is so important. The deviation from
the correct number is so unbelievably tiny that it represents an
insignificantly small amount of distortion. And what we're
really talking about here is distortion, no? Since all audio
ends up coming out of a loudspeaker eventually, by how many
micrometers (nanometers?) will the tweeter displacement be in
error when a DAW rounds 0.5 to 4.99999999999999? By how much
off will it be when an analog mixer has a whack at it? More
important, since most speakers have distortion in the single
digits anyway, that will swamp the tiny amounts we're talking
about for either analog or digital summing.
--Ethan
Ethan Winer
> It's not purely the act of adding--either in analog or digital
domains--that determines the "purity" of results. It's the
artifacts before, during and after the process that will
determine the sound. There are right ways and wrong ways to add
signals together in either domain. <
Please explain a little more. What artifacts? What do they
have to do with summing? What is the "wrong" way to combine
signals digitally?
Here are a few quotes from Roger Nichols a few months back in
his MusicPlayer forum. Sorry for the length, but these are too
good not to share.
--Ethan
===========================
May 2, 2001
It is amazing to me that nobody ever complained about analog
recording like they do about digital recording. I never heard
"The guard band is too narrow on the 3M machine, so I prefer the
Ampex MM-1000 for my recording. The flutter frequency is too low
on the MCI machine, and the distance between the erase head and
record head are too big to make tight punches. I like the 40
track head stack on the Stephens machine, but I will only mix to
a Scully with dbx."
Mostly you didn't hear about these things because you could not
measure them easily, or other analog anomalies masked them. Now
that we are in the digital era, there are new things to complain
about. The way to figure out if a system works well is to use
it, or listen to something that was done on it and see if you
like it.
I am doing a project right now completely in ProTools
24bit/48kHz. The musicians were great, everything sounds great,
so that is all I care about. The TDM buss doesn't sound thin to
me. What is a thin TDM buss supposed to sound like? I have done
a dozen albums completely in ProTools, including three Grammy
winning Bela Fleck albums.
You take what you have to work with and you make it sound good.
If you have an analog board to mix through like a Neve or SSL,
then that is what you use. The album of the year this year and
best engineered recording were recorded 16-bit and mixed through
an 30 year old analog Neve console. Analog consoles soften up
the sound a little. We liked it better through a digital
console, but we could not get the time we needed at a digital
console studio, so we mixed it through the analog console and
did the best we could. We could not have mixed it on an 02R or
d8b because of the fader resolution, not because of the digital
sound of the console. To make 0.1dB level changes you need 1024
step faders. Those consoles don't have that.
I have mixed other projects that were less exacting through the
02R and the d8b. They came out fine. I prefer the additional
benefits of digital all the way. The guys complaining about thin
TDM busses I will bet do not have a ProTools TDM system. There
are not 200,000 TDM systems out there because they have thin TDM
busses.
===========================
May 3, 2001
I knew someone who blew up the engine in her BMW just going back
and forth to work 5 miles each day. She was always in city
traffic and never exceeded the speed limits. It turns out that
the dealer sold her a standard transmission which she did not
know how to drive. She never took the car out of first gear,
running the engine near red-line all the time. It finally threw
a rod through the side of the engine block.
There is nothing wrong with the mix buss, it is the way you use
it.
I got a mastering job a few weeks ago. The guys worked a year on
their album in ProTools, mixed it in ProTools, and sent me the
files to master. It was the worst sounding piece of crap I have
ever heard in my 40 years of recording. You could not tell the
guitars from the keyboards, the bass from the vocals, the drums
from the distortion. This was not due to the mix buss. It was
just bad recording, bad mixing and bad effects perfectly
preserved in 24bit audio.
Analog recording masks tons of problems. The harmonic distortion
on analog tape is 100 to 1000 times more than digital, and helps
cover up some recording errors. Analog tape compresses the
signal if it is too loud, digital just clips it off. I have
heard hundreds of bad analog recordings. What you do need for
reference is to know what the original sounded like before it
was recorded. The only way that is possible is to be there
during the recording. If the instruments sound like crap, the
digital recording will preserve it better than the analog
recording.
As far as the mix buss thing, how come it works fine for me and
not for someone else? Maybe my recording levels are better.
Maybe you don't need all 64 tracks at full level with the faders
at zero mixing together with the master fader pulled down 40dB.
Gain structure must be followed in digital as well as analog. If
you are overloading the master fader, you pull down the source
tracks until you can put the master fader back up to zero. If
you don't do that then you will be distorting the mix buss on a
digital mixer AND an analog mixer. With no EQ, pan and level on
an 02R, d8b, DMX, and ProTools will be exactly the same. I have
listened to it, measured it, and make records that way every
day. End of discussion.
Maybe who ever started the rant has a bad console, or a bad TDM
buss cable, or some other broken device that is leading to these
conclusions. Until someone shows me some facts or collected data
to prove the claims, I am going to keep on working, thinking
that everything is alright.
Roger
Mark McQuilken
>
> Mark,
>
> > It's not purely the act of adding--either in analog or digital
> domains--that determines the "purity" of results. It's the
> artifacts before, during and after the process that will
> determine the sound. There are right ways and wrong ways to add
> signals together in either domain. <
>
> Please explain a little more. What artifacts? What do they
> have to do with summing? What is the "wrong" way to combine
> signals digitally?
>
> Here are a few quotes from Roger Nichols a few months back in
> his MusicPlayer forum. Sorry for the length, but these are too
> good not to share.
>
> --Ethan
>
>
> Roger
Ethan:
I'm not debating one way or the other about the supposed sound quality
of either approach, so presumably, Roger's quotes are directed more at
the rest of contributions to this thread (which I've not read).
My comment was about strictly in response to the idea that digital
summation is inherently better --this is my interpretation of your
comment-- 'cause, I believe, it ignores the system-level issues. Let's
say, for the sake of argument, that you've identified an addition
function that's "perfect" (it's interesting in this context that there
are a wide variety of "add" functions available to us depending upon the
CPU/architectural choices). As long as we operate this addition in
isolation, everything's cool. However, an adder that doesn't do anything
beyond the confines of an ALU isn't much use. Soooo, we've got to add
something from the outside world and get the result back to the outside
world. The artifacts that I mentioned, are (not exhaustive...just the
tip 'o' the iceburg):
* Less-than-perfect performance of the A/D process and the new signals
it will create or discount,
* The choice of "add" operation (fixed point--bit depth, floating point,
saturation 'rithmetic, etc.)
* The choice of data reduction from the additive process (i.e., add
things of large magnitude too many times and you run out of dynamic
range)
* The less-than-perfect performance of the D/A process and what it does
to our "perfect" addition
The correct addition must be selected for the rest of the system.
These are no different in structure to many of the issues in the analog
domain, they're just different specific system shortcomings. In the
analog domain, instead of ADC effects, we've got input/bias
currents/noise etc. (list of many, many, many error sources here), that
contribute not-necessarily-wanted errors to the summation. Add too many
things together and you run out of headroom, etc.
An "ideal" digital summer occurs about as frequently as an "ideal"
analog one. The system error sources may have different characteristics,
but both analog and digital summers cannot ignore how they are effected
*AND* how they effect the system-level outcome...
Sorry for the pithy explanation, but I hope that this better explains
what I meant by "artifacts".
Cheers,
McQ
Mark Plancke
>It is amazing to me that nobody ever complained about analog
>recording like they do about digital recording. I never heard
>"The guard band is too narrow on the 3M machine, so I prefer the
>Ampex MM-1000 for my recording. The flutter frequency is too low
>on the MCI machine, and the distance between the erase head and
>record head are too big to make tight punches.
I'll try to quote Fletcher here, the MCI JH24 punches like a
motherfucker. He'll correct me if I got it wrong, hopefully.
As far as your other non-head of events, perhaps every ones to elated
at the sound of the playback to worry about those things. Only half
kidding.
Mark Plancke
SOUNDTECH RECORDING STUDIOS
Windsor, Ontario, Canada
http://SoundTechRecording.com
Right now the recording game is crawling with wannabees. Lots of
profiteering can be done at their expense, by making flimsy crap and
pitching it as "just as good as, or a suitable alternative to" something
that's really good and has stood the test of time. - Steve Albini
Brian Tankersley
comments along the lines of "It is amazing to me that nobody ever complained
about analog recording like they do about digital recording." Huh?
Back when it was an analog only world, everybody with ears I ever knew spent
mucho time debating which analog multitrack sounded better for a given use
(Swedien used an MCI 16 track for drums and a Studer A800 for vocals), which
analog console to use for genre "x", etc. Major preferences from which tape
type and alignment particulars to which 2" machine and which mods done on
it, to custom built consoles using only 990s (or whatever) because the
prefab consoles didn't cut it for that guy, to the brand and type of wire
and patchbay size, to the particulars of the grounding scheme, etc, etc,
etc. Those preferences could only be based upon complaints about the gear
that was not preferred, and were often extremely technical in nature. Just
like today.
People avoided machines with bad head spacing or crappy ramping/monitoring
that made punching suck. Some machines could get out on a sustained piano
track and some could not. Again, very technical issues that created
complaints. Just like today. Fairly exotic issues like the *precise* bipolar
voltage supplied for a given model opamp or transistor. People could hear
the difference and prefered one or another voltage. Conceptually similar to
the tweak most PT users have discovered regarding using a USD or Aardsync WC
and avoiding the internal clock on an 888. Also similar to how people "in
the know" knew the stock summing bus on a Series 80 would not tolerate being
pushed *at all* while you could hammer the summing bus on a Neve. Many
modded their Tridents. I could go on and on.
Digital in no way whatsoever began, nor has a corner on the "tweakhead"
debating factor. People hear music differently and as a result, prefer
different equipment. And any specific piece of gear tends to inherently
steer you in a given direction as you use it based upon it's layout, it's
design philosophy, whatever coloration it lends, and a host of other subtle
cues. An 1176 screams "spank me" at you while a GML EQ says "Just a minute
sir, your table is almost ready".
Maybe some people complain about mixing in Pro Tools because the mixer
screen is so utterly bland looking that it sucks the vibe right out of the
air. I don't know. I do know that when I use different gear, I get somewhat
different results, all else being equal, for reasons ranging from the
technical to mental to the emotional. So does just about everybody else. If
Roger does not, he's one of a very few.
Regards,
Brian T
Mark Plancke wrote:
> "Ethan Winer" <nos...@nospam.com> wrote:
>
> >c I never heard
Marti D.Humphrey
"Mark Plancke" <Ma...@Soundtechrecording.com> wrote in message >
> I'll try to quote Fletcher here, the MCI JH24 punches like a
> motherfucker. He'll correct me if I got it wrong, hopefully.
>
> As far as your other non-head of events, perhaps every ones to elated
> at the sound of the playback to worry about those things. Only half
> kidding.
>
> Mark Plancke
> SOUNDTECH RECORDING STUDIOS
> Windsor, Ontario, Canada
> http://SoundTechRecording.com
>
Just a little story about analog machines . I over the years have had the
pleasure (at times) of working with the following
24 track machines. First the Ampex MM1200. This machine sounded wonderful
yet punches like hell. It took hours to set up the PERC cards to get them to
punch even close. Then you go ahead and rebias for new stock and the machine
needs to be tweaked again. But man they do make a great playback machine.
Back in 1982 I was working at Complete Post in Hollywood. I was assisting Ed
Greene and we were mixing Richard Harris in "Camelot" for HBO. We were
finally finishing the mix after about a month of very long days. We had gone
right down to the wire and Ed couldn't finish mixing the show at Complete
since there was another client in and they wouldn't let us run over into
their time. Doug Nelson another fine mixer was brought in to finish the show
and we took it to United Western across the street. This was before it
became "Ocean Way". Well they had a JH 24 MCI there and we finished up
mixing after a very long day at United Western but they didn't have a Video
Machine to layback to so I brought the 2 5000ft reels back to Complete and
started laying them back to the
one inch Video Master and all was great until we got to the area where we
were mixing on the MCI, every single punch was heard that we made at United
Western. We never heard any bad punches there but when we played back on the
Ampex 1200 we heard every single one. It seems that with the MCI head stack
there is both an azimuth and zenith adjustment. If for any reason you are
out of adjustment you will hear it on the 1200. I've worked with MCI's also
over the years and I wouldn't call them the best punching machine by any
means. They are OK, but I find the Sony 24 to punch much better and the
Otari
MTR 90-2 and especially the MTR 100 even better. The thing with the MCI's
are the "works in a drawer" cards for alignment. As soon as you slide them
in or out they seem to change. Replacing the Molex connections seem to help
but man is that a finicky machine. Give me a Tascam MMR-8 if you want great
punching all day long, in the middle of words or in music, nothing comes
close!!!
Now back to your machines,
Marti D. Humphrey C.A.S.
aka dr.sound
Monte P McGuire
Ethan Winer <nos...@nospam.com> wrote:
>I really don't see why that is so important. The deviation from
>the correct number is so unbelievably tiny that it represents an
>insignificantly small amount of distortion. And what we're
>really talking about here is distortion, no? Since all audio
>ends up coming out of a loudspeaker eventually, by how many
>micrometers (nanometers?) will the tweeter displacement be in
>error when a DAW rounds 0.5 to 4.99999999999999? By how much
>off will it be when an analog mixer has a whack at it? More
>important, since most speakers have distortion in the single
>digits anyway, that will swamp the tiny amounts we're talking
>about for either analog or digital summing.
How can you determine a priori that some "really small" amount of
distortion is inaudible? Have you figured out something that the rest
of us haven't? ;-)
We'd all like these problems to go away, but reality doesn't seem to
listen to logic like that. Low level problems do exist with low
resolution digital processors, and they tend to go away with
ridiculously overspec'ed processors. The "numbers' look very very
good in either case, yet they sound different. I don't know why they
sound different, but why should I "believe" that I can't hear
artifacts that I suspect might be at some ridiculously low level when
a) I can hear them, b) they might not actually be that low in level
when a processor is handling complex music signals and c) nobody is
testing the audibility of this stuff with real music signals.
Regards,
Monte McGuire
mcg...@world.std.com
John Klett
> We'd all like these problems to go away, but reality doesn't seem to
> listen to logic like that. Low level problems do exist with low
> resolution digital processors, and they tend to go away with
> ridiculously overspec'ed processors. The "numbers' look very very
> good in either case, yet they sound different. I don't know why they
> sound different, but why should I "believe" that I can't hear
> artifacts that I suspect might be at some ridiculously low level when
> a) I can hear them, b) they might not actually be that low in level
> when a processor is handling complex music signals and c) nobody is
> testing the audibility of this stuff with real music signals.
The sound quality degradation issue in ProTools and other DAW's comes up
time after time both in conversation and as part of my day to day work
trying to get studio systems working up to their potential, sorting out
specific problems with client systems and, sometimes, in the few
instances when I engineer.
You can argue this from theory... numbers is numbers, math is math and
so on so there CAN'T be a problem. But if this were the case there
would BE no complaint and no repeated statements by users and clients
who plainly feel that while there are many largely economic and
functional benefits to working in virtual studios they are certain they
are paying the price in other ways - mainly in loss of sonic integrity.
My view is that most people who argue the "numbers is numbers and math
is math" position are not listening at all or have not spent enough time
listening to good analog systems. As far as "math is math" goes
consider this... When you add up a bunch of numbers that are expressed
in 24 bits you will very often get a bigger number that can't be
expressed in 24 bits. That is okay - we have a 32 or 64 bit processor
so no "clipping" occurs. Further manipulation is required to reduce the
"gain" or otherwise get the resultant output back to 24 bit at some
point. In order to do that some information will have to be lost. This
is a rounding error. The easiest way to round a 32 bit result down to
24 bit is simply to truncate the result back to 24 bits. This requires
no processing and more quickly frees up the DSP or host processor to do
other things - and there are MANY other things to do. A better way to
deal with the necessity to round wide-bit math results back down to 24
bits is to DITHER... which requires more processing and reduces the
availability of the DSP or host processor to do other things.
The trend in product development and manufacturing for analog and
digital products geared to the mass market is to provide the most
features for the least price. In DAW-land DSP and high horsepower host
processors cost money. You will find that people who develop software
and plugs for these systems MUST work within a budget - a processor
overhead budget. They are given some number of MIPS or FLOPS in which
to accomplish a given function. How else can you run multiple plugs on
every virtual track and generate "aux sends" from those tracks to plugs
and so on and then mix all of this down without needing a full rack of
DSP? Here we have the digital version of TANSTAAFL.
What instructions and operations can you cut out and still accomplish
the plug within the processing budget? One thing the eats processor
overhead is dealing properly with rounding errors. So you deal with
rounding in a half-assed way or you simply abandon it and go with
truncation. In modeled systems you reduce the complexity of the
model... so maybe you drop the code that describes some aspect of
transformer core saturation or roughly approximates the varied transfer
characteristics of some particular remote cutoff variable mu tube
instead of detailing it out with more code that will blow the budget.
In these cases a code writer is making decisions that compromise the
potential for maintaining sonic integrity.
Digital audio can be very very good. Weiss and Z-Systems make very good
sounding digital EQ and Dynamics processors for mastering. These units
were conceived, designed and built with sonic integrity as the number
one goal. The code is well written and takes good care when reduction
processes occur in the math. The power supplies and clocking are well
done so jitter and other potential error sources are reduced. They are
made for audio. These are not mass market items. It is not that the
mass market would not really love gear that sounds good but the mass
market is not going to spend the money. So instead what the mass market
gets is stuff that gives tons of features and lots of tracks and the
ability to run lots of plugs - often for the same money as, say, one
Z-Systems EQ and one Weiss Dynamics Processor. These two boxes will
sound better than any of the multitude of things in the mass market
system but they can only do what they do and only to six channels.
At the end of the day I compare good analog to good digital and bad
analog to bad digital. I view ProTools in about the same light as an
AMEK BIG and a 1"16 track. You can make records on both of these
systems - many successful records have been made on these systems. I
would choose neither of these systems to work on. If I had to I could
probably get passable results from either system.
The biggest problem I have is that ProTools has become a standard.
Whether you like it or not you must have a ProTools system. If the Amek
BIG became the required standard analog console for all recording
studios to the extent that ProTools has become a standard for DAW's Amek
would certainly be happy but I am not sure everyone using them would be.
I generally campaign against bad audio and promote good audio. That
means I am not a ProTools fan. On the other hand it appears that you
have to have it. The fact that something is a standard does not mean it
is any good. Something becomes a standard because it does a passable
enough job offering a set of features in one product that people feel
they need and will spend money for in the absence of anything else they
perceive as a choice - add to this some large marketing budgets, tight
control over third party "partners", a beefy legal staff and some GREAT
timing and you have what amounts to be the equivalent of Windows in a DAW.
Take a look at Merging Technologies PyraMix and some of the other DAW's
out there. I and a few clients are evaluating that system and coming
away impressed (so far... I'll have a firm opinion after I do an
upcoming mix project on it). I have heard a number of mastering
engineers comment that stuff coming off Paris systems sounds more
musical - look at Paris. What is it about the Paris system that some
mastering engineers like? There are other systems that sound better
than ProTools. Ask yourself WHY and get on DigiDesign's case more about
sonics and less about features.
In general I judge audio systems with my ears. I can tell you if I
don't like the sound of something fairly quickly. I think the only
console that SSL makes that sounds GOOD is the 9K. I have always not
liked the sound of the E and G series - though G was an improvement over
E. The Euphonix 2000 and 3000 analog console have very good "numbers"
but they give me a sonic headache and wear me out. I don't know why
this is but something about the sound of those consoles hurts my ears.
Neve 80 Series consoles and the V Series sound good to me - but very
different from each other.... and so on.
I tried on several occasions to mix projects within ProTools and I found
the whole process totally unrewarding. No matter what I did everything
was dimensionally flat and sort of mashed together sounding. I gave up
and in one case where there was an analog console and enough outputs
from ProTools I brought out the individual tracks and mixed in the
analog console using analog outboard gear. This told me that ProTools
is a compromised environment and the less I did in there the better the
results COULD be. This has been born out time and again working with
various clients.
ProTools will be "upgrading" to 96/24 or better sometime next year.
Will be be 96/24 done well? I doubt it.
Unfortunately - this is my suspicious and cynical side talking - a large
percentage of ProTools users have no real experiance using good analog
gear or good digital gear in a way where they could learn what good
sound really is. Almost every record project today is somehow touched
by ProTools. The fact that records are still selling despite obvious
(to me at least) digital processing artifacts and over processing in
general does not make me believe that ProTools is any good....
...but I am aware that I am in the minority. I go for good audio. I
don't care how many people tell me that ProTools and other DAW's don't
mess with the sound in a bad way. I don't like how these systems sound.
I am NOT an "analog to the end" guy. I am not a "tubes are better than
transistors" guy. I don't believe that I am a fossil... though
unfortunately (or maybe fortunately) at some point we will ALL be fossils.
I also think that Microsoft BLOWS and I am and have been a Mac user
since 1985. I run Windows on my Mac only when I need to use an
application I can't get for MacOS... and it runs pretty quick too and
oddly enough does not crash on my mac... but I still think of it as an
unfortunately necessary evil.
sorry I took up so much space - have fun - flame away
--
John Klett / Technical Audio Engineering, Consulting & Design
http://www.technicalaudio.com/
Ethan Winer
> How can you determine a priori that some "really small" amount
of distortion is inaudible? Have you figured out something that
the rest of us haven't? ;-) <
Maybe <gg>. Seriously, distortion is not difficult to
understand. The two kinds of distortion that are most audible
are IM because it creates inharmonic components, and anything
that generates excessive upper partials that are far enough away
from the fundamental that they are not masked. I'll also
include with IM any aliasing problems caused by inadequate
filters because those too are inharmonic.
I really don't think small amounts of distortion (> 40+ dB.
below the signal, or less than 1%) are as big a problem as many
people think. If it were, nobody would like the sound of analog
tape!
> nobody is testing the audibility of this stuff with real music
signals. <
I'm sure lots of people test with real music. I certainly use
only music for listening tests, and never test tones. But
unless a test is performed double blind, or the gear being
tested is grossly inadequate, it's all but useless.
--Ethan
Mark Plancke
>
>"Mark Plancke" <Ma...@Soundtechrecording.com> wrote in message >
>> I'll try to quote Fletcher here, the MCI JH24 punches like a
>> motherfucker. He'll correct me if I got it wrong, hopefully.
>>
>> As far as your other non-head of events, perhaps every ones to elated
>> at the sound of the playback to worry about those things. Only half
>> kidding.
>>
>> Mark Plancke
>> SOUNDTECH RECORDING STUDIOS
>> Windsor, Ontario, Canada
>> http://SoundTechRecording.com
>>
>MTR 90-2 and especially the MTR 100 even better. The thing with the MCI's
>are the "works in a drawer" cards for alignment. As soon as you slide them
>in or out they seem to change. Replacing the Molex connections seem to help
>but man is that a finicky machine.
Actually my JH24 came out of a post production studio (Masters
Workshop) in Toronto. They had modified this and their other JH24's to
be able to tweak the ramp up and down times. Apparently they could
punch syllables without a problem. I don't have those cards here and
I'm still impressed with how well the JH24 punches. Mine's a '87
w/ALIII so I can't speak for any other revisions of this machine. As
far as stability goes, this one is a champ, I'm down to doing a MRL
playback alignment monthly without having to tweak very much at all.
Mark Plancke
SOUNDTECH RECORDING STUDIOS
Windsor, Ontario, Canada
http://SoundTechRecording.com
Right now the recording game is crawling with wannabees. Lots of
Ethan Winer
> My comment was about strictly in response to the idea that
digital summation is inherently better <
Actually, my point was merely that digital summing is not
inherently bad. The original question asked about digital
summing, and implied that it sucks because it seems like the
only way to get good sound is to go out of the DAW to an
external mixer. This is just silly, which is why I posted
Roger's common sense quotes.
> (it's interesting in this context that there are a wide
variety of "add" functions available to us depending upon the
CPU/architectural choices). <
Actually, it's not how values are added that varies, it's how
the data is stored: the number of bits and the size of mantissa
and exponent. If you have two 32-bit floating point numbers in
IEEE format, they'll combine to give the same result regardless
of the CPU or even the algorithm, no?
> As long as we operate this addition in isolation ...
Less-than-perfect performance of the A/D process and <
But summing IS done in isolation! You have two or twelve tracks
being played back, and you add the numbers for the current
sample after any plug-in processing. A/D/A conversion or EQ or
whatever is outside that loop, and has nothing to do with the
accuracy of the addition itself.
> These are no different in structure to many of the issues in
the analog domain, they're just different specific system
shortcomings. <
Agreed.
--Ethan
Mark McQuilken
>
>
> Actually, it's not how values are added that varies, it's how
> the data is stored: the number of bits and the size of mantissa
> and exponent. If you have two 32-bit floating point numbers in
> IEEE format, they'll combine to give the same result regardless
> of the CPU or even the algorithm, no?
Unless Intel makes the CPU???? <g>
>
> > As long as we operate this addition in isolation ...
> Less-than-perfect performance of the A/D process and <
>
> But summing IS done in isolation!
It's only one piece in many required processes--I'm speaking strictly
here--and has /no/ value unless it is combined with other processes
(like I/O). For example (sorry if you know this already), one very
viable and common design approach for opamp stages is to assume an ideal
opamp and /then/ go back to account for error sources. Although it's a
groovy (and effective) way to do it, it'd be patently wrong to assume
that 'cause of this "perfect" opamp model, the final circuit will behave
that way. Same with the digital summer. Even if a given "add" command is
perfect (like our "perfect" opamp model), it's strengths and foibles
will only be realized in application...
> You have two or twelve tracks
> being played back, and you add the numbers for the current
> sample after any plug-in processing. A/D/A conversion or EQ or
> whatever is outside that loop, and has nothing to do with the
> accuracy of the addition itself.
I'm not talking about the accuracy (or lack thereof) of the addition. I
made an earlier point of saying "assume a perfect addition" (in reality
it doesn't exist...as you know *ANY* binary system is finite set
arithmetic. There will /always/ be a resolution issue.)...a summing
stage is not only about addition (analog or digital). A "perfect" summer
that, when fit into the rest of the system, does harm to signal
integrity is a badly implemented process no matter how perfect the
addition is (for example, a system that deals with logarithmic signals
could yield undesired results by the "perfect" addition...unless you
want to modulate the signals) . Here's my point: even if we agree that a
particular addition function is "perfect", a DAW's success or failure as
a summer is not entirely (or even largely) dependent the CPU's
performance of that addition. Just because the summer is operating in
the digital domain with a level of "perfection" in that domain, says
nothing about the DAWs operation as a summing buss in the same way that
an accurate, very low noise opamp doesn't guarantee that it's
implementation as a summer is going to be low noise or accurate. Both a
perfect "add" instruction or perfect opamp are useless /in vacuo/...
>
> > These are no different in structure to many of the issues in
> the analog domain, they're just different specific system
> shortcomings. <
>
> Agreed.
>
> --Ethan
Cheers,
McQ
hank alrich
> I don't believe that I am a fossil... though
> unfortunately (or maybe fortunately) at some point we will ALL be fossils.
Only the great ones will be fossilized; the rest of us will become
topsoil, eventually.
> sorry I took up so much space - have fun - flame away
Take all the space you need, John; much of that rant ought to be in the
FAQ!
--
hank alrich * secret__mountain
audio recording * music production * sound reinforcement
"If laughter is the best medicine let's take a double dose"
hank alrich
> I really don't think small amounts of distortion (> 40+ dB.
> below the signal, or less than 1%) are as big a problem as many
> people think. If it were, nobody would like the sound of analog
> tape!
Since what we can measure and what we hear seem to be at odds, perhaps
there is more to the characterization of "distortion" than mere
quantification of artifacts without reference to their quality. I feel
safe saying that since what we don't know vastly exceeds what we do know
(and nevermind what we think we know <g>), we will eventually learn
enough to understand why certain audio reproduction discrepancies are
irritating and other not.
hank alrich
> If Roger does not, he's one of a very few.
Hey, he said there was no reason not to get a Mackie d8b, and that was
when the out of sync L and R master outs situation was still roaming
bugland.
Kevin F. Rose
the post could remain as is. Nice job, I can sleep now.
Rick Krizman
LilYellowMonstr wrote:
>
>
> Digital mixing seems to be too complex, in that the summing buss issue is
> really frustrating. For instance, I sat in with a guy yesterday who was mixing
> nuendo tracks. He routes right out his 9652 card into a little Spirit Digital
> board. The two mix is then recorded to two seperate tracks on the computer. I
> asked him, why don't you just mix entirely within Nuendo, then he did this,
> and we compared the mixes and the one summed through the 328 was undeniably
> better, more space, more 3D. Yikes. Its just too much. I wonder if anyone
> has tried doing this with Pro Tools and another DIGITAL mixer.
>
Yes, somebody did this comparing Protools to a Sony Oxford and found the results
to be identical in every respect. AFAIK this was a legitimate and valid test and
I've seen no reason put forth to doubt the results.
RIck Krizman
KrizManic Music
Rick Krizman
Mark McQuilken wrote:
> An "ideal" digital summer occurs about as frequently as an "ideal"
> analog one. The system error sources may have different characteristics,
> but both analog and digital summers cannot ignore how they are effected
> *AND* how they effect the system-level outcome...
>
> Sorry for the pithy explanation, but I hope that this better explains
> what I meant by "artifacts".
The question remains, however, as to how to best improve the digital mixing
bus. Do we try to make it more accurate or do we find a way to make it
less accurate by modelling all the euphonically desireable distortion that
an analog console provides? I suspect the latter will sound better.
Rick Krizman
KrizManic Music
Rick Krizman
John Klett wrote:
> Take a look at Merging Technologies PyraMix and some of the other DAW's
> out there. I and a few clients are evaluating that system and coming
> away impressed (so far... I'll have a firm opinion after I do an
> upcoming mix project on it). I have heard a number of mastering
> engineers comment that stuff coming off Paris systems sounds more
> musical - look at Paris. What is it about the Paris system that some
> mastering engineers like? There are other systems that sound better
> than ProTools. Ask yourself WHY and get on DigiDesign's case more about
> sonics and less about features.
Fine, but exactly what are we supposed to tell Digi? Most of these
discussions put the blame, I think erroneously, on the accuracy of the
summing bus. I don't think making a more "accurate" summing bus will address
any of the problems you described. I think that what you're missing is
analog distortion to glue it all together. There needs to be a plugin that
models this at every stage of the mix bus. I think a good name for the
plugin would be FORGIVENESS.
Rick Krizman
KrizManic Music
LS1productions
bus. >>
And here is where I say, "if you want to make the digital summing better, what
is a reference of this 'better' ". And the answer is a good analog console, a
Neve, an SSL, an API, an Amek... Just as Universal audio modeled the LA-2A
and the 1176 to such great success, a channelstrip from a Neve, an SSL J, a
Legacy, etc. should all be modeled part by part. Then go into the summing amps
in these consoles. Model that. Model the trip from the channel strip TO the
summing amp. Now put it all together and you have modeled a real analog
console, and if it is done as well as UA did those comps, we can all rest in
ease. Put the DSP on some dedicated cards, and bammm.
<< Do we try to make it more accurate >>
Fuck accuracy. If we should all learn something by all this, it is that
accuracey sucks, 456 and a Neve rules. Lets concentrate more on the ladder.
LS
Ethan Winer
> I feel safe saying that since what we don't know vastly
exceeds what we do know ... <
I really doubt we know as little as many people seem to believe.
Pro audio is not rocket science by any stretch.
What's really needed to sort this stuff out is more double-blind
testing. I think a lot of folks would be surprised at how
similar most pro audio equipment is. I truly believe in the
placebo effect, the "it cost a lot so I know it's good" (and
vice versa) effect, and the power of suggestion.
Even if trained ears can discern tiny differences between one
[pick literally anything] and another, that doesn't mean it's
the limiting factor in the quality of a recording. In my
experience the real defining factors are the musicians, their
instruments, the recording space, and to a lesser extent the
microphones used. Too many people believe you can't produce a
hit unless you have [pick literally anything once again].
--Ethan
Ethan Winer
> Unless Intel makes the CPU???? <g>
Ouch! :->)
> it'd be patently wrong to assume that 'cause of this "perfect"
opamp model, the final circuit will behave that way. Same with
the digital summer. Even if a given "add" command is perfect
(like our "perfect" opamp model), it's strengths and foibles
will only be realized in application... <
I can't agree with that. Yes, op-amps are imperfect. (I've
designed plenty of op-amp circuits in my day.) But how is CPU
addition ever not perfect?
> a summing stage is not only about addition
Agreed. When all those tracks combine to exceed 0 dB., you then
have to add a minus offset to push it back down to less than 0.
But that too is just more addition...
> Just because the summer is operating in the digital domain
with a level of "perfection" in that domain, says nothing about
the DAWs operation as a summing buss in the same way that an
accurate, very low noise opamp doesn't guarantee that it's
implementation as a summer is going to be low noise or accurate.
<
Okay, then let me ask you this: In what way can a DAW's summing
buss be inadequate? What would account for the complaints we
hear about summing busses? It drives me nuts when people talk
about audio degradation (analog, digital, or from not using
Monster cable <g>) in terms of losing "sound stage" or
dimension, or any of the things that I know for a fact are due
more to stereo arrival time than cable losses through
capactitance. Maybe they really are hearing something, but they
sure ain't describing it very well!
--Ethan
Ethan Winer
> ... or do we find a way to make it less accurate by modelling
all the euphonically desireable distortion that an analog
console provides? I suspect the latter will sound better. <
No, please, no! The goal must always be to strive for accuracy.
If YOU want to cloud up your mixes with distortion - even
desirable distortion - please leave me out of it. I'm content
to copy a mix to a cassette and back or use a plug-in to add
that effect.
--Ethan
Mark McQuilken
It's clear that I'm not making myself clear and, unfortunately, I'm too
busy to write more than a few missives. Just a quick couple of points,
then I'll sign off (if you'd like to talk about this, I'll be happy to
discuss by phone):
Ethan Winer wrote:
>
>
> I can't agree with that. Yes, op-amps are imperfect. (I've
> designed plenty of op-amp circuits in my day.) But how is CPU
> addition ever not perfect?
Let me see if this makes sense: (1) View the "perfect" opamp model
(assume for a moment that it actually exists) as the core for a larger
process, say, summation. (2) Obviously, there's more to making a summer
than just this "perfect" opamp model. (3) There are error sources
throughout. (4) The final summer is the "perfect" opamp combined with
the error sources. You with me so far? Agree? Yes? No? Now, take items
(1) - (4) and substitute "CPU addition" everywhere I've listed "opamp".
Same form, different specific implementation. Is the summer more or less
perfect with either the ideal opamp or ideal CPU addition? It depends.
What does it depend upon? Peripheral error sources and how the
analog/digital additions effect the output.
>
> > a summing stage is not only about addition
>
> Agreed. When all those tracks combine to exceed 0 dB., you then
> have to add a minus offset to push it back down to less than 0.
> But that too is just more addition...
Although division can be implemented using multiple subtractions
(archaic), I wouldn't put it quite that way. What you're talking about
is scaling, usually implemented with fractional multiplication (FMUL) or
some DIV operand. Now, just for the sake of argument, use your
floating-point additions (is that the most "ideal" in your opinion?) on
16, 20 or 24-bit audio (gotta convert one to the other). Let's also
assume that the equipment user has max'ed out the input levels to, say,
16 channels (i.e., each input is occuring at 0dBFS). Use your "perfect"
CPU addition to combine them into a single output *AND* make sure the
output fits into the 16, 20 or 24-bit output format *AND* do it without
producing /any/ artifacts. Can't be done. Doesn't matter how perfect
your addition is. Remember, I'm not even arguing the accuracy of your
mythically perfect CPU addition (finite set arithmetic is *NOT* infinite
set arithmetic, although for many practical systems it's good enough).
I'm willing to say, for practical purposes, that a given addition is
good enough. However, (again, I say), addition /en vacuo/ is not a
summing buss.
>
>
> Okay, then let me ask you this: In what way can a DAW's summing
> buss be inadequate?
Too many ways to be covered here. There's plenty of great material on
the subject. Call me if you'd like some references.
> What would account for the complaints we
> hear about summing busses? It drives me nuts when people talk
> about audio degradation (analog, digital, or from not using
> Monster cable <g>) in terms of losing "sound stage" or
> dimension, or any of the things that I know for a fact are due
> more to stereo arrival time than cable losses through
> capactitance. Maybe they really are hearing something, but they
> sure ain't describing it very well!
>
> --Ethan
Good talking with you...
McQ
FMR Audio
512-280-6557
hank alrich
> a DAW's success or failure as a summer
Only those who spend the big bucks get a summer DAW; the rest of us live
with the winter versions.
Ethan Winer
> I'm too busy to write more than a few missives. <
I understand, and I appreciate the time you've taken to discuss
this so far!
> The final summer is the "perfect" opamp combined with the
error sources. ... Now, take items (1) - (4) and substitute "CPU
addition" everywhere I've listed "opamp". <
But this is the very crux of the matter. Op-amps have known
failings - less than infinite gain, higher than theoretical
minimum noise floor, etc. ALL of the "failings" of digital
summing are due ONLY to math rounding errors.
> What you're talking about is scaling, usually implemented with
fractional multiplication (FMUL) or some DIV operand. <
Yes, you're absolutely right that the final scaling is
multiplication and not addition or subtraction.
> Use your "perfect" CPU addition to combine them into a single
output *AND* make sure the output fits into the 16, 20 or 24-bit
output format *AND* do it without producing /any/ artifacts.
Can't be done. Doesn't matter how perfect your addition is. <
Again you are correct. It's impossible to combine all those
numbers and end up with a PERFECT result. But I contend that
you CAN achieve well under 0.001% distortion, and to me that's
mighty transparent. Surely good enough to not be the limiting
factor in an audio signal chain.
>> Okay, then let me ask you this: In what way can a DAW's
summing buss be inadequate? <<
> Too many ways to be covered here. <
I'd be happy with one or two ways <smile>.
--Ethan
Bob Olhsson
<kl...@technicalaudio.com> wrote:
>My view is that most people who argue the "numbers is numbers and math
>is math" position are not listening at all or have not spent enough time
>listening to good analog systems.
Don't forget that bad math is also bad math!
--
Bob Olhsson Audio Mastery Recording Project Design and Consulting
Box 90412, Nashville TN 37209 Tracking, Mixing and Mastering
615.352.7635 FAX 615.356.2483 Mix Evaluation and Quality Control
40 years of making people sound better than they thought possible!
Rick Krizman
Ethan Winer wrote:
> Rick,
>
> > ... or do we find a way to make it less accurate by modelling
> all the euphonically desireable distortion that an analog
> console provides? I suspect the latter will sound better. <
>
> No, please, no! The goal must always be to strive for accuracy.
Does your mic cabinet reflect that philosophy? Ever stick a 57 on a
snare? Ever use a compressor?
>
> If YOU want to cloud up your mixes with distortion - even
> desirable distortion - please leave me out of it. I'm content
> to copy a mix to a cassette and back or use a plug-in to add
> that effect.
>
I don't think it's a choice between an accurate mix and a muddy
cassette. Nor does there exist a plugin that can truly emulate the
sound of a great analog console. We're slicing it a little finer here.
Rick Krizman
KrizManic Music
Kevin Smith
Thanks, John, and Monte too.
"John Klett" <kl...@technicalaudio.com> wrote in message
news:3B8266C4...@technicalaudio.com...
Ethan Winer
> Does your mic cabinet reflect that philosophy? Ever stick a
57 on a snare? Ever use a compressor? <
Good points <smile>. Actually, I hate 57s and never use them,
mostly because they're so peaky and lack an extended high end.
I also avoid compression on most things, though not all. I
record a lot of classical music, so accuracy is important there.
Then again I agree with your philosophy, and even with classical
music a clarity and pleasing sound are more important than
accuracy.
But in your original comment you talked about improving digital
mixing and asked "Do we try to make it more accurate or do we
find a way to make it less accurate by modelling all the
euphonically desireable distortion that an analog console
provides?" I can's see how purposely losing accuracy or adding
distortion is desireable in a mixer. I can see using any effect
you find beneficial - and there's no reason a plug-in can't be
designed to do that - but somehow "less accurate" goes against
the grain for me.
--Ethan
Rick Krizman
Ethan Winer wrote:
> But in your original comment you talked about improving digital
> mixing and asked "Do we try to make it more accurate or do we
> find a way to make it less accurate by modelling all the
> euphonically desireable distortion that an analog console
> provides?" I can's see how purposely losing accuracy or adding
> distortion is desireable in a mixer. I can see using any effect
> you find beneficial - and there's no reason a plug-in can't be
> designed to do that - but somehow "less accurate" goes against
> the grain for me.
You can't have it both ways. If you have an accurate sound and change
it, it's going to be less accurate. That whole analog euphonic thang is
distortion.
Rick Krizman
KrizManic Music
Robert Orban
Actually, a great deal _is_ known about how much distortion is audible
in the presence of a given signal. In fact, there is now an ITU standard
describing an algorithm for estimating this. (The algorithm was
originally designed for the objective testing of perceptual coders and
tested against results of ABC/hr subjective listening tests.)
If one wants to set a criterion for audibility, I would consider the
unmasked threshold of hearing to be a ridiculously conservative metric
for estimating the the audibility of added distortion, and am confident
that distortion below the unmasked threshold of hearing will not be
mysteriously heard in the presence of a masker.
for example, see:
Audio Engineering Society
Preprint Index Entry
Title: Can Objective Methods Replace Subjective Assessments?
Author: Christer Grewin
Year: 1995
Preprint No: 4067
Convention No: 99
Abstract: Currently the only reliable method for evaluating the audio
quality of low bit-rate audio codecs is by subjective assessments. This
is a costly and very time-consuming procedure. Furthermore, existing
coding standards are very flexible and allow for large differences in
implementation without violating these standards. Thus, evaluations of
individual implementations are necessary, but cost and efforts may
hamper this. It is desirable that subjective assessments can be
replaced, or at least complemented, by an objective measurement method.
A number of proposals for objective perceptual models have been
described in literature. At present, these methods are not sufficiently
validated. In 1994, ITU-R established Task Group 10-4 to recommend a
method for objective perceptual measurements of audio equipment. A draft
recommendation is expected by the end of 1996. This paper reports on the
work in ITU-R Task Group 10-4 and the requirements as outlined by the
group. Experiences gained by the Swedish Broadcasting Corporation when
using one of the proposed models, PAQM from PTT Netherlands, are also
discussed.
Audio Engineering Society
Journal Index Entry
Title: Evaluating a Measurement System
Authors: T. Sporer; U. Gbur; J. Herre; R. Kapust
Year: 1995
Volume: 43
Issue: 5
Page: 353-363
Abstract: Noise-to-mask ratio (NMR) is a perceptual measurement scheme
which gives information about the distance between actual noise and
masking threshold. It has been shown to be a useful tool in the
development and comparison of perceptual coding schemes. Some
"perceptual experiments" carried out with the measurement system as a
"test subject" are presented. The results of these measurements are
compared with the results obtained with human listeners.
ŠAudio Engineering Society, Inc.
John Klett
> Fine, but exactly what are we supposed to tell Digi? Most of these
> discussions put the blame, I think erroneously, on the accuracy of the
> summing bus. I don't think making a more "accurate" summing bus will address
> any of the problems you described. I think that what you're missing is
> analog distortion to glue it all together. There needs to be a plugin that
> models this at every stage of the mix bus. I think a good name for the
> plugin would be FORGIVENESS.
You are right - analog distortion is generally harmonic and somehow
musically relative to the program. Many years ago I sat and watched Roy
Thomas Baker spend a day or two at the top of a project just running
tracks through a console and distorting different parts to see how he
could use it to glue the recording and mix together. That convinced me
that analog distortion can be good. Look at how much effort has gone
into modeling analog distortion in digital devices (the new version
Cranesong HEDD is a great thing by the way) - but artifacts that result
from just about ANY digital error you can name are not in anyway
musically related to the program material so, while musical distortion
CAN be useful there is no use that I can think of for non-musical
distortion. I don't think a FORGIVENESS plug would fix the math and
remove bad digital artifacts.
Since math and rounding errors make non-musical artifacts that we can
perceive (that we generally don't like) I would think that the object
should be to fix the math. Doing the math the right way requires more
processing power. I don't believe there is anything inherently bad
about digital done well. The summing busses and most other processes in
Oxford, Capricorn, System 5, Axiom MT and so on sound a lot better to me
than those in ProTools. One difference is that instead of writing code
to fit a processing budget so that more features and plugs can run in
the limited space available the large format console manufacturers went
for enough dedicated processing power to accomplish better math in all
the processes the console makes available. Large format digital console
have less problems with sonic integrity though they are certainly not
problem free.
The only thing I can think of to say to Digi is to make more processing
power available so that code writers can have bigger processing budgets
to work with. The problem here is that since the vast majority of
ProTools users don't hear a problem or have consciously decided that
they can make the sonic sacrifice in order to work more productively
Digi-Design has very little incentive to fix anything. The marketing
decision that has made Digi look towards 96/24 is not a response to
people asking for better audio quality. 96/24 is just a feature that
people have been asking for - like many others - and it has to run on a
system designed and built to a fit very tight budget.
Mike Rivers
> Fine, but exactly what are we supposed to tell Digi?
Tell them that they're supplying a system for a "world class" facility
that's ready to spend $200,000 or whatever it takes to get really
"world class" audio. No tight budget, no problem.
But Digidesign knows that they can sell 100 $20,000 systems for every
one $200,000 system. So you establish a new facility standard
workstation. This industry is very fickle.
Just one comment on John Klett's excellent overview - while high
horsepower host processors may be expensive, this is not the problem
of the software designer. Everyone who hasn't upgraded a computer
because he wanted (or needed) to run a new piece of software please
turn in your mouse. Once you've spent the big bucks on the software
and made the commitment to the learning curve, the hardware will get
bought. Othewise you will have wasted a lot of money you can't get
refunded.
--
I'm really Mike Rivers (mri...@d-and-d.com)
LS1productions
remove bad digital artifacts.
>>
Thats what I've been thinking lately too. Putting on a McDsp analog channel on
each track of a PT Mix won't take away the fact that the system is not as deep
and wide as a good analog console. I tried doing this the other day, and while
there was a thickness that I had never acheived from an all PT mix, it still
lacked the 3D and space of the mix I did the following day, sending the same
tracks out to a console.
Paul Winkler
tangent - I realize now it was the old analog-vs-digital thing rather
than the old digital-vs-digital thing.
Having said that I'd like to clarify some ambiguous things in my
earlier post, and then I'll try to shut up and get out of this one.
On Tue, 21 Aug 2001 01:05:25 GMT, Ethan Winer <nos...@nospam.com>
wrote:
>Paul,
> >> BUT as a sometime computer programmer I can tell
you that it's >simply not the case that all addition is equivalent. <
>I've been a professional programmer for 20 years, and I don't
>know of very many different ways to add 2 plus 2.
OK, let me try to be clearer. I was trying to communicate that 2.0 +
2.0 is not the same as 2 + 2. It may be the same operator but the
numbers are represented quite differently.
> I agree
>integer math can have rounding errors during many computations.
>But I was talking about summing, not signal processing.
OK, maybe I'm stretching by referring to summing as "signal
processing." Yes, there are no rounding errors in integer addition. I
should have made that clear. Change gain by a non-integer amount and
it's a different story: we're now multiplying, probably discarding a
fractional portion of the result. This is pretty much inevitable - I'd
guess not too many mixes end up with all faders _exactly_ on zero, all
EQ bypassed, no processing whatsoever.
>Using
>8-byte double-precision values (64 bits) gets you VERY close to
>ideal for the relatively modest precison digital audio requires.
I don't doubt it! Even 32-bit floats carry quite a lot of
information. I did not mean to imply otherwise.
>> In this case, 1.0 / 2.0 = 0.5, or (this is important)
>something PRETTY CLOSE to 0.5, like maybe 0.49999999999999999.
>The disadvantage here is that there's a slight inaccuracy right
>off the bat - many integers simply cannot be perfectly
>represented as a floating point number. <
>
>I really don't see why that is so important. The deviation from
>the correct number is so unbelievably tiny that it represents an
>insignificantly small amount of distortion.
I meant that it is important for understanding the difference between
integers and floats. Whether such a small inaccuracy is audible is
another question entirely, and I will not opine on that; I have no
basis for doing so one way or the other.
Just wanted to add my limited programmer's perspective... hope it was
useful to somebody.
--PW
Ethan Winer
> Having said that I'd like to clarify some ambiguous things in
my earlier post, and then I'll try to shut up and get out of
this one. <
There's no need to shut up. <smile> This is important stuff,
and if more people understood it properly there'd be a lot less
arguing over nonsense like whose $100 per foot speaker cables
sound better!
> 2.0 + 2.0 is not the same as 2 + 2. It may be the same
operator but the numbers are represented quite differently. <
Agreed, but either way you still get 4, or at least something
VERY close to that.
> Change gain by a non-integer amount and it's a different
story: we're now multiplying, probably discarding a fractional
portion of the result. <
Again I agree. What it really comes down to is how much
distortion will that create? 0.01%? 0.001%? 0.0000001%? I
maintain that any distortion products more than, say, 40-60 dB.
below the signal are pretty well inaudible, unless they are
"unusual" products such as aliasing whistles, etc., not what
you'd get from simple math errors.
--Ethan
Mark McQuilken
>
(snip)
> portion of the result. <
>
> Again I agree. What it really comes down to is how much
> distortion will that create? 0.01%? 0.001%? 0.0000001%? I
> maintain that any distortion products more than, say, 40-60 dB.
> below the signal are pretty well inaudible, unless they are
> "unusual" products such as aliasing whistles, etc., not what
> you'd get from simple math errors.
>
> --Ethan
Ethan:
We're all entitled to our opinion, particularly with such subjective
stuff (that's *NOT* intended as a snide comment). I don't think that the
technical literature bears your position out. Talk with Rupert sometime
about it. He's made quite a study of it, has conferred with many
renowned researchers about this and his conclusions are very different
from yours. I've even seen evidence with my test gear/monitoring set-up
that Rupert's probably correct... As *COMPLETELY* anecdotal evidence:
why would the converter manufacturers spend million's ($$$) on reducing
distortion components by refining their products? This activity is *NOT*
done strictly in a measurement lab...they've undertaken--in the
development of the converters--to more strongly correlate the distortion
products with the "impressions" of the golden-eared people. Design and
development organizations tend to be incredibly pragmatic (particularly
the ones driven by engineering managers). If your assertion were
perceived by everybody as correct, think of the money they'd save!
McQ
Rick Krizman
John Klett wrote:
>
>
> Since math and rounding errors make non-musical artifacts that we can
> perceive (that we generally don't like) I would think that the object
> should be to fix the math.
If you're referring to Protools, I'm not sure it has been demonstrated that any
math or rounding errors are at all audible. That doesn't mean Protools sounds as
good as an analog console--it just points to a possible different reason than the
undocumented cliche that Protools math has a "problem".
> Doing the math the right way requires more
> processing power. I don't believe there is anything inherently bad
> about digital done well. The summing busses and most other processes in
> Oxford, Capricorn, System 5, Axiom MT and so on sound a lot better to me
> than those in ProTools.
In a well known recent experiment, the exact same mix was run through Protools and
an Oxford with identical results. (they nulled out) You really think all those
summing busses sound different? What is it that informs your opinion?
> The only thing I can think of to say to Digi is to make more processing
> power available so that code writers can have bigger processing budgets
> to work with.
I think that's a misdirected effort--where is any evidence to support this
position? What is needed are some sound files from different digital systems to
which you can point and say "See how much better this one is than Protools", then
figure out why.
> The problem here is that since the vast majority of
> ProTools users don't hear a problem or have consciously decided that
> they can make the sonic sacrifice in order to work more productively
> Digi-Design has very little incentive to fix anything.
I'm sure if you could demonstrate to Digi that there was a problem with their math
then they would have an incentive to fix it. AFAIK, there is no evidence anywhere
to support this.
Rick Krizman
KrizManic Music
Rick Krizman
LS1productions wrote:
I'm not a big fan of Analog Channel. To my ears it provides the worst of analog
without giving that deep and wide feeling.
Rick Krizman
Krizmanic Music
Rick Krizman
Mike Rivers wrote:
> In article <3B823A8C...@mediaone.net> rkri...@mediaone.net writes:
>
> > Fine, but exactly what are we supposed to tell Digi?
>
> Tell them that they're supplying a system for a "world class" facility
> that's ready to spend $200,000 or whatever it takes to get really
> "world class" audio. No tight budget, no problem.
>
No, my question was "what are we supposed to tell Digi to do to improve
their product?" Out of all this anecdotal bashing of the Protools summing
bus and all these claims that such-and-such workstation has "better
sounding" math, nothing has been audibly demonstrated. I too would like to
see the product sound "better", and I have a subjective sense of what I mean
by "better", but I think applying vastly greater DSP resources to get a more
accurate summing bus will not accomplish this.
I anxiously await any shred of evidence to the contrary.
Rick Krizman
KrizManic Music
name withheld
Rick Krizman <rkri...@mediaone.net> wrote:
> I'm not a big fan of Analog Channel. To my ears it provides the worst of analog
> without giving that deep and wide feeling.
>
> Rick Krizman
> Krizmanic Music
>
i agree... its a good simulation of overdriving the record head and
badly adjusted bias, but it doesnt do much of the "good" things analog
tape does like smoothness and depth.
--
aaron
+37dB studios
san bruno, CA
irc:muf^n icq#56140852
Richard Novak
ls1prod...@aol.com wrote on 8/22/01 8:52 AM:
> ... a PT Mix ....is not as deep
> and wide as a good analog console. I tried doing this the other day, and
>(the PT mix) still
> lacked the 3D and space of the mix I did the following day, sending the same
> tracks out to a console.
Pretty interesting. Was there no difference between the mixes other than
the analogue summing vs the PT summing?
I've been thinking that the ideal way to mix from ProTools would be in
8-Track stems; Kick, Snare, Bass, Lead Vocal, (in Mono) plus two stereo
pairs of everything else, recorded onto a 1" Studer. (Assuming your
mastering guy also had a 1" 8-Track.)
I work Post on "Young & the Restless" and that's how we get the audio;
as stems on DA-88, ch 1 & 2 mono, ch 3 - 8 = 3 stereo pairs. Set the tones
to "0" and you've got your mix.
Cheers, Rick Novak.
Monte P McGuire
Ethan Winer <nos...@nospam.com> wrote:
>Again I agree. What it really comes down to is how much
>distortion will that create? 0.01%? 0.001%? 0.0000001%? I
>maintain that any distortion products more than, say, 40-60 dB.
>below the signal are pretty well inaudible,
I think that's overly generous. But, you're welcome to your own
opinions. If they match up with the world as you hear it, then great!
>unless they are
>"unusual" products such as aliasing whistles, etc., not what
>you'd get from simple math errors.
OK, so what happens when 10KHz intermodulates with 15KHz in a 44.1KHz
sampled system? You get a 5KHz difference product and a 25KHz sum
product that aliases down to 19.1KHz. Aliasing is not a tough thing
to make happen in a low sample rate system. Consider this and what
happens inside of a digital compressor...!
My final point is that you should look at the spectral content of the
trash produced by truncation. It is not at all simple or euphonic
like the harmonic distortion caused by a long tailed transistor pair
(for example). It's not so easily masked because there's energy in a
lot of places.
Regards,
Monte McGuire
mcg...@world.std.com
Monte P McGuire
LS1productions <ls1prod...@aol.com> wrote:
>Fuck accuracy. If we should all learn something by all this, it is that
>accuracey sucks, 456 and a Neve rules. Lets concentrate more on the ladder.
Just as a historical anecdote, all of the gear that is now coveted as
vintage gear was designed with the goal of accuracy in mind. They
were not going for euphonic distortion. Of course, they had these
real world components to deal with and they ended up with some
distortion and coloration, but they got to where they did not by
trying to make "colored" gear, but by trying to eliminate the errors.
Given how misunderstood and poorly applied dither is, I contend that
we still aren't making digital gear that's accurate enough to worry
about adding some dirt to the chain. The concept of "self dither" is
something that almost every digital chain relies on, and that's silly.
Right at the start of the chain, the decimation filter in a sigma
delta converter is the first place where the signal is truncated
without dither. Most processing paths give you a few more further
down in the chain. I can't see how anyone could say that we've run
the full course with dither and shown that it isn't something that
could help a lot when nobody is listenbing to a chain free of
undithered truncations.
Regards,
Monte McGuire
mcg...@world.std.com
Kurt Ballou
> I've been thinking that the ideal way to mix from ProTools would be in
> 8-Track stems; Kick, Snare, Bass, Lead Vocal, (in Mono) plus two stereo
> pairs of everything else, recorded onto a 1" Studer. (Assuming your
> mastering guy also had a 1" 8-Track.)
That's fine for TV production, but do you really want to give up your
ability to mix to the mastering guy? What about the way that the
tracks interact, particularly with a bit of mix compression? As the
mixer, you should maintain control of the mix down to the final
format. If you are mixing 5.1, then sure, go to 1" 8 track, but for
stereo mixes, why leave it up to the mastering guy?
LS1productions
>accuracey sucks, 456 and a Neve rules. Lets concentrate more on the ladder.
>>
What I really should have said was this: why is it that so many eq and
compressor and delay plugins are all modeled after analog hardware pieces, but
there are no digital mixers modeled after analog consoles? Why can't someone
model the distortion characteristics of a Neve 9098 channel and then how
multiple channels combine in the 9098 summing buss, just like they modeled
Pultec curves for the Waves Renn Eq, and just how they modeled the components
of an 1176 for the UAD-1 Dsp Card. Rather then just adding the bits or
whatever, why not make the bits sound like something that is known to be good?
Kevin Smith
helps, but still sounds like a computer to me. I've worked with 16-48
channels of stems and the more the merrier, to a point. If every stem
already has gain changes or processing the digital footprint is harder to
work around. Any d/a anamolies tend to pile up too, so in those cases I
think you're actually better off with fewer channels.
But if it's treated like a multitrack, with little to no (the less the
merrier) gain changes, processing, or submixing at the computer, things
sound a lot more normal to me, whatever that is. Less computer footprint.
The drawback is that the whole mix has to built up from scratch in the
expensive studio. Basically, it comes down to budget - the more work you do
in the computer at the cheap studio, the faster you can get the job done at
the good one, but the less you gain sonically.
Kevin Smith
residual, just extremely low in level. I do remember that all the fx were
pre-mixed on one of the systems then transferred to the other. Necessary
for their test, but low level, ambient detail is one of the main problems I
have with the Pro Tools mixer. If they were pre-mixed on the Oxford then
brought into the DAW at unity gain, or vice verse, we aren't comparing their
relative performance at mixing ambient effects.
Anyway, I've seen the Digi guys allude to DSP bandwidth as being a
limitation on the DUC, so I really think there's something to it. If the
mixer used 3 Mix cards just to run itself it might sound a lot better, but
then they'd have a marketing problem...
Mixerman
"Ethan Winer" <nos...@nospam.com> wrote in message
Yawn.
You don't know what you're talking about. And Roger's in the severe
minority when it comes to rather well established names in recording, and
their thoughts on this particular subject.
Mixerman
Arny Krueger
> > The final summer is the "perfect" opamp combined with the
> error sources. ... Now, take items (1) - (4) and substitute "CPU
> addition" everywhere I've listed "opamp". <
This is very flawed logic. It basically is like saying:
(1) op amps add things imperfectly.
(2) Computer accumulators also add things.
(3) Therefore since both op amps and computer accumulators add
things, and op amps are subject to imperfections, computer
accumulators must be subject to the same imperfections.
Mathematics is a very abstract environment in which total perfection
and utter predictability are possible.
Real world op amps are neither as perfectible nor as predictable.
> But this is the very crux of the matter. Op-amps have known
> failings - less than infinite gain, higher than theoretical
> minimum noise floor, etc. ALL of the "failings" of digital
> summing are due ONLY to math rounding errors.
The various kinds of errors that may happen in some digital adders
can be reduced to any desired small amount by the simple addition of
appropriate hardware.
No amount of added hardware or effort can possibly make an
operational amplifier achieve arbitrary performance levels.
Once an op amp achieves a noise figure of 1 dB, its noise can never
be improved by more than 1 dB.
Digital adders that maintain a 144 dB dynamic range with audio data
are cheap and common, while analog audio circuits that have 144 dB
are generally unobtainable.
> > Use your "perfect" CPU addition to combine them into a single
> output *AND* make sure the output fits into the 16, 20 or 24-bit
> output format *AND* do it without producing /any/ artifacts.
However the artifacts can be reduced to any arbitrary level by simply
throwing more hardware at the problem. You can't do that with op
amps.
> Can't be done. Doesn't matter how perfect your addition is. <
Digital addition can be as perfect as one desires. Any computer that
is generally programmable can be programmed with algorithms that will
perform common mathematical operations to any desired level of
accuracy, even to a million digits.
> Again you are correct. It's impossible to combine all those
> numbers and end up with a PERFECT result. But I contend that
> you CAN achieve well under 0.001% distortion, and to me that's
> mighty transparent. Surely good enough to not be the limiting
> factor in an audio signal chain.
Anybody who worries about 0.001% or less nonlinear distortion needs
to spend some time listening to
http://www.pcabx.com/technical/nonlinear/index.htm .
> >> Okay, then let me ask you this: In what way can a DAW's summing
buss be inadequate? <<
> > Too many ways to be covered here. <
> I'd be happy with one or two ways <smile>.
The real point is that some given DAW implementation can have a
summing bus that is inadequate. However, there is nothing inherent in
digital arithmetic that is responsible for this. OTOH analog circuits
can only be so good, and no amount of time or money can possibly
improve them beyond certain easy-to-predict performance levels
established by thermal noise.
Mark McQuilken
I've responded to some specific things below and disagreed with many of
your points. However, as you'll see, the disagreements are based upon
what I believe to be out-of-context interpretations. I don't think, in
the final analysis, we're far apart on much of this...
Arny Krueger wrote:
>
> "Ethan Winer" <nos...@nospam.com> wrote in message
> news:SAAg7.96270$ai2.6...@bin2.nnrp.aus1.giganews.com...
>
> > > The final summer is the "perfect" opamp combined with the
> > error sources. ... Now, take items (1) - (4) and substitute "CPU
> > addition" everywhere I've listed "opamp". <
This is my quote, not Ethan's, *AND* you're:
(a) taking it out of context, and,
(b) completely missing the point.
Here it is in distilled form:
An digital audio summing bus is not merely the "perfect" addition of two
audio samples. There is nothing trivial about a well-designed digital
audio summer. There are other processes required to get the samples
in/out (this conversation was deliberately limited to exclude hardware
I/O). There are significant, measurable errors that occur as a result of
these other processes (going under the assumption.../for the sake of
argument/, not reality...that our addition is "perfect"). The
system-view of this process has a direct analog (excuse the pun): the
design of an analog summer. One successful design method is to assume
that the opamp is "perfect" and account for errors through the
application of this "perfect" element to the outside world. This design
technique not only works analytically, it also works in the real-world.
Now to your points:
>
> This is very flawed logic.
I don't agree, based upon the context of the discussion.
>It basically is like saying:
>
> (1) op amps add things imperfectly.
I didn't say that. If the distilled explanation above is not clear, wade
through the rest of the thread...
>
> (2) Computer accumulators also add things.
Lost me there, Arny.
>
> (3) Therefore since both op amps and computer accumulators add
> things, and op amps are subject to imperfections, computer
> accumulators must be subject to the same imperfections.
Again, you need to read in the context. I stated *multiple* times that
where the analogy diverges is in the /particular/ nature of the error
sources.
>
> Mathematics is a very abstract environment in which total perfection
> and utter predictability are possible.
I won't even go there. Mathematics /can/ be abstract. It can also be
pure and applied. We benefit daily from a world of daily conveniences
realized through the gate of mathematics.
>
> Real world op amps are neither as perfectible nor as predictable.
Never said they were. But they are predictable to within a given error
bound.
The following is one of Ethan's quotes:
> > But this is the very crux of the matter. Op-amps have known
> > failings - less than infinite gain, higher than theoretical
> > minimum noise floor, etc. ALL of the "failings" of digital
> > summing are due ONLY to math rounding errors.
>
> The various kinds of errors that may happen in some digital adders
> can be reduced to any desired small amount by the simple addition of
> appropriate hardware.
Only within the constraints of the selected implementation techniques.
>
> No amount of added hardware or effort can possibly make an
> operational amplifier achieve arbitrary performance levels.
I may be taking this last sentence out-of-context or not really
understand what your trying to say, but if I take it at face value, it's
just wrong. What I think you meant is that there is some performance
extrema over which, an opamp cannot be made to perform through any
external means (example: OP176 /maximum/ noise performance is determined
by external components/interface techniques--my point BTW--but it
*CANNOT* be made any quieter by external components than it's internal
implementation will allow). *ALL* system implementations have
performance extrema, artifact constraints and, in the real world,
non-technical constraints. The art and science of design is to balance
the selection and implementation of the constituent elements to achieve
as-close-to-desired performance as is possible within those constraints.
>
> Once an op amp achieves a noise figure of 1 dB, its noise can never
> be improved by more than 1 dB.
Ah! So you that's *WAS* what you were saying.
>
> Digital adders that maintain a 144 dB dynamic range with audio data
> are cheap and common, while analog audio circuits that have 144 dB
> are generally unobtainable.
Nothing could be further from the truth! I'm sitting at my test bench
now, with a non-esoteric amplifier stage (i.e., it's got very few parts
and they're all off-the-shelf) with a dynamic range of 142dB (noise
floor = -112dBu, clip point = +30dBu). Given a few more parts, I could
easily have a stage operating beyond 144dB in a matter of hours. I've
got a good friend--a very good analog designer--who has designed
slightly more complicated stages that FAR EXCEED 144dB. The 144dB DR
target is *NOT* that difficult to hit in the analog realm...
What I believe you're talking about is noise floor, *NOT* strict dynamic
range. In /that/ context, of course, there's an inherent, practical
limitation...but there's nothing that says we can't build upwards!
>
> > > Use your "perfect" CPU addition to combine them into a single
> > output *AND* make sure the output fits into the 16, 20 or 24-bit
> > output format *AND* do it without producing /any/ artifacts.
>
> However the artifacts can be reduced to any arbitrary level by simply
> throwing more hardware at the problem.
Not within the system constraints and the context of this discussion.
Going from a "perfect" n-bit floating point internal architecture to an
external 16-bit I/O structure--*NOT* including the hardware...strictly
data sets--will not improve the inherent limitations of the 16-bit
format. To quote Zolzer, although well-known by many DSP practitioners:
"It can be noticed that the signal-to-noise ratio for fixed-point
representation depends on the input level. This signal-to-noise ratio in
the digital domain is an exact image of the level-dependent
signal-to-noise ratio of an analog signal in the analog domain. A
floating-point representation cannot improve this signal-to-noise ratio.
Rather, the floating-point curve is vertically shifted downwards to the
value of the signal-to-noise ratio of an analog signal." This
limitation, BTW, is analogous to the thermal noise floor of the opamp...
>You can't do that with op
> amps.
>
> > Can't be done. Doesn't matter how perfect your addition is. <
>
> Digital addition can be as perfect as one desires. Any computer that
> is generally programmable can be programmed with algorithms that will
> perform common mathematical operations to any desired level of
> accuracy, even to a million digits.
If we're gonna pick nits, then "perfect as one desires" just ain't
correct. Perfection is independent of one's desire. Finite-set
arithmetic, although it can *approach* theoretical perfection, will
*ALWAYS* fall short of the infinite-set case. It can achieve--to use
your choice of words--an "arbitrary" performance level. Pick an error
bound and then construct the arithmetic to get you under that error
bound. However, with finite-set arithmetic, no matter how many bits in
your mantissa/exponent, I can always find a "desired" error that is less
than that.
Having said that, this part of the discussion is not an exercise in
practicality. Getting back to context of the previous part of the
thread:
An digital audio summing bus is not merely the "perfect" addition of two
audio samples. There is nothing trivial about a well-designed digital
audio summer. There are other processes required to get the samples
in/out. There are significant, measurable errors that occur as a result
of these other processes (going under the assumption.../for the sake of
argument/, not reality...that our addition is "perfect").
>
> > Again you are correct. It's impossible to combine all those
> > numbers and end up with a PERFECT result. But I contend that
> > you CAN achieve well under 0.001% distortion, and to me that's
> > mighty transparent. Surely good enough to not be the limiting
> > factor in an audio signal chain.
>
> Anybody who worries about 0.001% or less nonlinear distortion needs
> to spend some time listening to
> http://www.pcabx.com/technical/nonlinear/index.htm .
Well, since I know we can't even begin to agree about this, I won't
discuss it. We'll agree to disagree.
>
> > >> Okay, then let me ask you this: In what way can a DAW's summing
> buss be inadequate? <<
>
> > > Too many ways to be covered here. <
>
> > I'd be happy with one or two ways <smile>.
>
> The real point is that some given DAW implementation can have a
> summing bus that is inadequate.
Thanks! That's my point. There's nothing magical about the DAW
implementation platforms that guarantee that "all digital summing busses
are the same" or "perfect".
>However, there is nothing inherent in
> digital arithmetic that is responsible for this.
This is where we, again, disagree. This is a practical matter. Although
we could design a 20-bit, highly linear audio system (including I/O), it
wouldn't be compatible with our current 16-bit delivery system...with
it's attendent limitations. These limitations *ARE* determined by the
16-bit fixed-point arithmetic.
>OTOH analog circuits
> can only be so good, and no amount of time or money can possibly
> improve them beyond certain easy-to-predict performance levels
> established by thermal noise.
Think outside the box, Arny! It's true that the low-level limitations
are themally limited (which, BTW, is the thing that KILLING converter
improvements), so we go the other direction! If you're limited to 0.5nV
per root Hz on the low end, then increase the operating voltages!
Regards,
McQ
WillStG
<< ...they've undertaken--in the development of the converters--to more
strongly correlate the distortion
products with the "impressions" of the golden-eared people. >>
Mark, I was gonna make a stupid joke, like - "golden eared, schmoel-den
eared. Let's go ahead and tell it like it is, people who have ears like dogs
- Dog eared! Yeah that's right, dog-eared people!"
But then I read your dissertation on this thread and now I'm like man,
but McQ's really a scientist and all.
Ah what the hell, "Aroooooouughhhhhh....." < '' >
Cheers,
Will Miho
NY Music & TV Audio Guy
Fox And Friends/Fox News
"The large print giveth and the small print taketh away..." Tom Waits
>>Mark McQuilken <A HREF="mailto:ma...@fmraudio.com">ma...@fmraudio.com</A>
wrote>>
Arny:
Now to your points:
Lost me there, Arny.
> <A
HREF="http://www.pcabx.com/technical/nonlinear/index.htm">http://www.pcabx
.com/technical/nonlinear/index.htm</A> .
Regards,
McQ<<
Will Miho
NY Music & TV Audio Guy
Fox And Friends/Fox News
"The large print giveth and the small print taketh away..." Tom Waits
hank alrich
> But then I read your dissertation on this thread and now I'm like man,
> but McQ's really a scientist and all.
Get real, man. He eats Mexican food and has an Ovation "guitar". That
last thing wasn't exactly designed by Rupert.
--
hank alrich * secret__mountain
audio recording * music production * sound reinforcement
"If laughter is the best medicine let's take a double dose"
Dave Martin
"hank alrich" <walk...@thegrid.net> wrote in message
news:1eypjly.1my...@209-162-27-70.thegrid.net...
: WillStG <wil...@aol.com> wrote:
:
: > But then I read your dissertation on this thread and now I'm like
man,
: > but McQ's really a scientist and all.
:
: Get real, man. He eats Mexican food and has an Ovation "guitar". That
: last thing wasn't exactly designed by Rupert.
Well yeah, but a man who eats good Mexican food can't be too stupid... And
we actually have a new Mexican restaurant out here in my neck of the woods!
I've eaten there twice this week. That gives us a choice when I'm not in the
mood for Barbecue. It ain't Mi Tierras (San Antonio) or Mercado Juarez
(Dallas/Fort Worth area), but the hot sauce is good.
Joelton Tennessee, the dining capitol of the world...
--
Dave Martin
DMA, Inc.
Nashville, TN
Harvey Gerst
>: >But then I read your dissertation on this thread and now I'm like man,
>: > but McQ's really a scientist and all.
>: Get real, man. He eats Mexican food and has an Ovation "guitar". That
>: last thing wasn't exactly designed by Rupert.
>Well yeah, but a man who eats good Mexican food can't be too stupid...
But that doesn't explain away the Ovation "guitar".
Harvey Gerst
Indian Trail Recording Studio
http://www.ITRstudio.com/
Dave Martin
"Harvey Gerst" <har...@ITRstudio.com> wrote in message
news:560A248664F0EBAE.D14109B5...@lp.airnews.net...
Oh, yeah. There is that. But look on the bright side - no one said that Mark
actually plays it...
hank alrich
> "Harvey Gerst" <har...@ITRstudio.com> wrote in message
> news:560A248664F0EBAE.D14109B5...@lp.airnews.net...
> : >"Dave Martin" <dave....@nashville.com> wrote:
> : >>"hank alrich" <walk...@thegrid.net> wrote:
> : >: WillStG <wil...@aol.com> wrote:
> : >: >But then I read your dissertation on this thread and now I'm like man,
> : >: > but McQ's really a scientist and all.
> : >: Get real, man. He eats Mexican food and has an Ovation "guitar". That
> : >: last thing wasn't exactly designed by Rupert.
> : >Well yeah, but a man who eats good Mexican food can't be too stupid...
> : But that doesn't explain away the Ovation "guitar".
> Oh, yeah. There is that. But look on the bright side - no one said that Mark
> actually plays it...
That's true; I never did hear him play it. It was just hanging on one of
his livingroom walls, begging not to be played. But what does this say
about the man if he thinks of something like _that_ as wall art? Maybe
it's an attempt at dysenterior design.
Roger W. Norman
guitar stand!
--
Roger W. Norman
www.SirMusicStudio.com
Ro...@SirMusicStudio.com
301-585-4681
"guys, it takes a lifetime to just get just a BIT closer..."
George Massenburg
"Harvey Gerst" <har...@ITRstudio.com> wrote in message
news:560A248664F0EBAE.D14109B5...@lp.airnews.net...
Dave Martin
:
: That's true; I never did hear him play it. It was just hanging on one of
: his livingroom walls, begging not to be played. But what does this say
: about the man if he thinks of something like _that_ as wall art? Maybe
: it's an attempt at dysenterior design.
I dunno - I think someone's gonna bring me a banjo to hang on a wall in the
studio, and one room is just screaming for a Velvet Elvis...
Do those things reflect badly on my taste?
Roger W. Norman
SOUND? Once you take all this hubbub out into the room via speakers, none
of that makes any difference if it doesn't sound right. That brings things
back around to being a subjective stance. I prefer to use my computer as a
recording device and mix via my analog console because I like the way it
SOUNDS. I mix in my computer because it's actually quicker although
somewhat less intuitive, but that doesn't mean that I can get the sound I
want by just putting on a tactile surface for the computer and mix somewhat
similarly to my analog console. But if I were to do so, and make virtually
all my mixing moves the same way, to me the sound would still be better via
my analog console. Unfortunately trying to do so in a controlled
environment where both mixes had the exact same moves and EQs, dynamics,
etc., isn't possible, so it still comes down to being a subjective decision.
And Ethan's contention that delivering the lowest noise floor equals
transparency is a wash to me. Unless one can take all sources from
beginning to end in the digital domain, there's got to be some coloration
occurring in the analog gain staging that's going to change the original
source. Now Ethan does this, composing and arranging his compositions
within the computer, so for his type of working (mostly, not always - he is
quite an accomplished musician who plays cello beautifully and also is an
excellent guitar player) he has the advantage of having somewhat a
transparent environment to work within. But once you place a mic into the
equation, there goes the transparency no matter what else takes place.
Whatever coloration is introduced in the gain stage is going to be amplified
in the math digitally. It's going to be changed, perhaps in suble ways, but
perhaps not (obviously the same goes for an analog console). If the sound
at the end of the entire process works for him, then it works for him. But
it has no bearing on whether it works for me in my work environment.
So let's just say, in relation to the sound of the final product, that a)
there's no objective method to test either contention, and; b) personal
preference will out. Once any subjectivity comes into play there's no way
to come up with a decisive and objective answer. The sound of the art
shouldn't be weighlayed by the technical aspects of "this vs that".
Personally, I don't care. Neither PT nor any console I've seen/worked on
has had my toes tapping. Nobody hums a console.
--
Roger W. Norman
www.SirMusicStudio.com
Ro...@SirMusicStudio.com
301-585-4681
"guys, it takes a lifetime to just get just a BIT closer..."
George Massenburg
"Arny Krueger" <ar...@hotpop.com> wrote in message
news:SCNh7.2493$Wz7.100...@newssvr16.news.prodigy.com...
Arny Krueger
"Mark McQuilken" <ma...@fmraudio.com> wrote in message
news:3B87CD...@fmraudio.com...
> Arny:
>
> I've responded to some specific things below and disagreed with
many of
> your points. However, as you'll see, the disagreements are based
upon
> what I believe to be out-of-context interpretations. I don't think,
in
> the final analysis, we're far apart on much of this...
>
> Arny Krueger wrote:
> >
> > "Ethan Winer" <nos...@nospam.com> wrote in message
> > news:SAAg7.96270$ai2.6...@bin2.nnrp.aus1.giganews.com...
> > > > The final summer is the "perfect" opamp combined with the
> > > error sources. ... Now, take items (1) - (4) and substitute
"CPU
> > > addition" everywhere I've listed "opamp". <
> This is my quote, not Ethan's, *AND* you're:
> (a) taking it out of context, and,
> (b) completely missing the point.
> Here it is in distilled form:
> An digital audio summing bus is not merely the "perfect" addition
of two
> audio samples. There is nothing trivial about a well-designed
digital
> audio summer.
At this point the discussion seems properly limited to the issue of
just summing audio signals.
> There are other processes required to get the samples
> in/out (this conversation was deliberately limited to exclude
hardware
> I/O).
At this point the discussion has been exploded to include the
universe, and then amended to contradict the explosion.
Net result: Meaninglessness and confusion.
> There are significant, measurable errors that occur as a result of
> these other processes (going under the assumption.../for the sake
of
> argument/, not reality...that our addition is "perfect").
Since these "other processes" aren't specified or designed, the whole
discussion becomes highly opaque.
>The
> system-view of this process has a direct analog (excuse the pun):
the
> design of an analog summer. One successful design method is to
assume
> that the opamp is "perfect" and account for errors through the
> application of this "perfect" element to the outside world. This
design
> technique not only works analytically, it also works in the
real-world.
Nice, but again, it sheds no light.
> Now to your points:
> >
> > This is very flawed logic.
> I don't agree, based upon the context of the discussion.
Which is now completely indecipherable.
> >It basically is like saying:
> > (1) op amps add things imperfectly.
> I didn't say that. If the distilled explanation above is not clear,
wade
> through the rest of the thread...
Yes, the explanation above is very unclear and highly
self-contradictory. No specifics at all, just a lot of vague claims.
> > (2) Computer accumulators also add things.
> Lost me there, Arny.
What could be clearer? Computer accumulators add things, don't they?
> > (3) Therefore since both op amps and computer accumulators add
> > things, and op amps are subject to imperfections, computer
> > accumulators must be subject to the same imperfections.
> Again, you need to read in the context. I stated *multiple* times
that
> where the analogy diverges is in the /particular/ nature of the
error
> sources.
But that's not the question I was addressing, or relevant to the
point I was making.
> > Mathematics is a very abstract environment in which total
perfection
> > and utter predictability are possible.
> I won't even go there. Mathematics /can/ be abstract.
You "went there" even though you said you weren't going to!
>It can also be pure and applied. We benefit daily from a world of
daily conveniences
> realized through the gate of mathematics.
More words that it was promised that they weren't going to be said.
> > Real world op amps are neither as perfectible nor as
predictable.
>Never said they were. But they are predictable to within a given
error bound.
Which unlike mathematical errors are dominated by random processes.
In short, real world op amps are neither as perfectible nor as
predictable as math. If I do the same arithmetic in the world of
mathematics, I get the same results to any desired degree of
precision. Can't say that about the world of analog electronics.
> The following is one of Ethan's quotes:
> > > But this is the very crux of the matter. Op-amps have known
> > > failings - less than infinite gain, higher than theoretical
> > > minimum noise floor, etc. ALL of the "failings" of digital
> > > summing are due ONLY to math rounding errors.
> > The various kinds of errors that may happen in some digital
adders
> > can be reduced to any desired small amount by the simple addition
of
> > appropriate hardware.
> Only within the constraints of the selected implementation
techniques.
Irrelevant because I didn't say "specific implemented techniques", I
said that the set of implementation techniques for digital arithmetic
is open-ended. Not enough precision? Just add gates, steps of
calculation, whatever.
> > No amount of added hardware or effort can possibly make an
> > operational amplifier achieve arbitrary performance levels.
> I may be taking this last sentence out-of-context or not really
> understand what your trying to say, but if I take it at face value,
it's
> just wrong.
Actually, you immediately contradict yourself. You take what I said
at face value and prove that I'm right.
> What I think you meant is that there is some performance
> extrema over which, an opamp cannot be made to perform through any
> external means (example: OP176 /maximum/ noise performance is
determined
> by external components/interface techniques--my point BTW--but it
> *CANNOT* be made any quieter by external components than it's
internal
> implementation will allow). *ALL* system implementations have
> performance extrema, artifact constraints and, in the real world,
> non-technical constraints.
All implementations have performance limits, but there are
performance limitations in the world of analog electronics that can't
be beaten. In the world of digital adders, you are doing digital
arithmetic. Off-the-shelf software can do digital arithmetic to any
desired level of precision, even hundreds and thousands of digits.
There is nothing like it in the analog world.
> The art and science of design is to balance
> the selection and implementation of the constituent elements to
achieve
> as-close-to-desired performance as is possible within those
constraints.
But, in the world of digital adders, multipliers and dividers, the
constraints are so large as to be well beyond ridiculous for audio
purposes. Not so for analog electronics. Not so for the actual
recording and playback environment.
> > Once an op amp achieves a noise figure of 1 dB, its noise can
never
> > be improved by more than 1 dB.
> Ah! So you that's *WAS* what you were saying.
And I said it.
> > Digital adders that maintain a 144 dB dynamic range with audio
data
> > are cheap and common, while analog audio circuits that have 144
dB
> > are generally unobtainable.
> Nothing could be further from the truth! I'm sitting at my test
bench
> now, with a non-esoteric amplifier stage (i.e., it's got very few
parts
> and they're all off-the-shelf) with a dynamic range of 142dB (noise
> floor = -112dBu, clip point = +30dBu).
Why did you miss my stated goal by 2 dB?
Your example is not responsive to my performance goal. It's just one
stage, not an entire working audio component. And, its a pretty
exceptional stage at that. Just guessing here, but I suspect that
this "stage" has a gain control in it, and the
claimed performance is not obtainable at a single setting of that
gain control. That's not the same as what I was talking about, which
is 144 dB dynamic without changing gain settings. Increasing dynamic
range by riding the gain is not the same thing as having a system
with a certain amount of dynamic range with a constant gain setting.
>Given a few more parts, I could easily have a stage operating
beyond 144dB in a matter of hours.
OK, let's up the ante from 24 bits to 32 bits. 32 bit digital
systems are common and have 192 dB of dynamic range.
> I've
> got a good friend--a very good analog designer--who has designed
> slightly more complicated stages that FAR EXCEED 144dB. The 144dB
DR
> target is *NOT* that difficult to hit in the analog realm...
Not as stated. Furthermore its totally irrelevant to the entire
analog recording/playback realm which includes room noise.
> What I believe you're talking about is noise floor, *NOT* strict
dynamic
> range. In /that/ context, of course, there's an inherent, practical
> limitation...but there's nothing that says we can't build upwards!
Sure there is a limit to building upwards. If you want low noise
transistors, you find that they have limited voltage range. AFAIK
there is no such thing as a truly low-noise transistor with a 1000
volt VCE limit. More like about 40 volts, right?
Then, in order to have low noise you need low circuit impedances. If
you build a mic preamp that puts out 100 volts rms into a 50 ohm load
(chosen for low noise floor), you have some heat dissipation issues,
right?
> > > Use your "perfect" CPU addition to combine them into a single
> > > output *AND* make sure the output fits into the 16, 20 or
24-bit
> > > output format *AND* do it without producing /any/ artifacts.
> > However the artifacts can be reduced to any arbitrary level by
simply
> > throwing more hardware at the problem.
> Not within the system constraints and the context of this
discussion.
I see no stated system constraints for the digital system.
> Going from a "perfect" n-bit floating point internal architecture
to an
> external 16-bit I/O structure--*NOT* including the
hardware...strictly
> data sets--will not improve the inherent limitations of the 16-bit
> format.
This is the year 2001 right? Why are we still talking about digital
systems that are limited to 16 bits?
16 bits is a distribution format for consumers, not a production
format for people who do audio production. It's an irrelevant limit
in the context of the practical limits of modern professional audio
production system.
> To quote Zolzer, although well-known by many DSP practitioners:
> "It can be noticed that the signal-to-noise ratio for fixed-point
> representation depends on the input level. This signal-to-noise
ratio in
> the digital domain is an exact image of the level-dependent
> signal-to-noise ratio of an analog signal in the analog domain. A
> floating-point representation cannot improve this signal-to-noise
ratio.
> Rather, the floating-point curve is vertically shifted downwards to
the
> value of the signal-to-noise ratio of an analog signal." This
> limitation, BTW, is analogous to the thermal noise floor of the
opamp...
However, the word length of a floating point system can be extended
to 100's of digits. Corresponding improvements in the analog domain
require liberal quantities of Unobtainium.
> >You can't do that with op amps.
> > > Can't be done. Doesn't matter how perfect your addition is. <
> > Digital addition can be as perfect as one desires. Any computer
that
> > is generally programmable can be programmed with algorithms that
will
> > perform common mathematical operations to any desired level of
> > accuracy, even to a million digits.
> If we're gonna pick nits, then "perfect as one desires" just ain't
correct.
It's not a nit. Its a practical fact. If I want to do the summation
of audio signals with 100 decimal digits of precision, its just a
matter of turning the crank.
> Perfection is independent of one's desire.
With digital arithmetic, ones desires can be totally ludicrous for
audio purposes, and they can be easily met.
>Finite-set
> arithmetic, although it can *approach* theoretical perfection, will
> *ALWAYS* fall short of the infinite-set case.
Right, but analog electronics falls shorter sooner. There's a reason
why analog computers are very rare. It has something to do with the
fact that analog electronics are not as easy to extend in the
direction of speed and accuracy as digital electronics.
>It can achieve--to use your choice of words--an "arbitrary"
performance level. Pick an error
> bound and then construct the arithmetic to get you under that error
> bound. However, with finite-set arithmetic, no matter how many bits
in
> your mantissa/exponent, I can always find a "desired" error that is
less
> than that.
Of course, but with analog, you run out of gas sooner, and end up
spending more money doing it.
> Having said that, this part of the discussion is not an exercise in
> practicality. Getting back to context of the previous part of the
> thread:
> An digital audio summing bus is not merely the "perfect" addition
of two
> audio samples. There is nothing trivial about a well-designed
digital
> audio summer. There are other processes required to get the samples
> in/out. There are significant, measurable errors that occur as a
result
> of these other processes (going under the assumption.../for the
sake of
> argument/, not reality...that our addition is "perfect").
Well, by taking the discussion way out of the bounds of just the
audio summer, you've found that there are other limits. That's a
classic way to "win" an argument, but its not fair play.
> > > Again you are correct. It's impossible to combine all those
> > > numbers and end up with a PERFECT result. But I contend that
> > > you CAN achieve well under 0.001% distortion, and to me that's
> > > mighty transparent. Surely good enough to not be the limiting
> > > factor in an audio signal chain.
> > Anybody who worries about 0.001% or less nonlinear distortion
needs
> > to spend some time listening to
> > http://www.pcabx.com/technical/nonlinear/index.htm .
> Well, since I know we can't even begin to agree about this, I won't
> discuss it. We'll agree to disagree.
That's right. By dealing with observable facts, I put you at a great
disadvantage.
> > > >> Okay, then let me ask you this: In what way can a DAW's
summing
> > buss be inadequate? <<
>
> > > > Too many ways to be covered here. <
>
> > > I'd be happy with one or two ways <smile>.
> > The real point is that some given DAW implementation can have a
> > summing bus that is inadequate.
> Thanks! That's my point. There's nothing magical about the DAW
> implementation platforms that guarantee that "all digital summing
busses
> are the same" or "perfect".
Those are absolutely ridiculous statements, of course there is
nothing magic about DAWs. However, they are far more perfectible, and
more perfectible to a reasonable degree for a lower cost.
> >However, there is nothing inherent in
> > digital arithmetic that is responsible for this.
> This is where we, again, disagree. This is a practical matter.
Although
> we could design a 20-bit, highly linear audio system (including
I/O), it
> wouldn't be compatible with our current 16-bit delivery
system...with
> it's attendant limitations. These limitations *ARE* determined by
the
> 16-bit fixed-point arithmetic.
This is yet another example of dragging a discussion outside its
natural limits.
> >OTOH analog circuits
> > can only be so good, and no amount of time or money can possibly
> > improve them beyond certain easy-to-predict performance levels
> > established by thermal noise.
> Think outside the box, Arny! It's true that the low-level
limitations
> are thermally limited (which, BTW, is the thing that KILLING
converter
> improvements), so we go the other direction! If you're limited to
0.5nV
> per root Hz on the low end, then increase the operating voltages!
But practically-speaking you can only increase the operating voltages
so far, and then other issues like the need to maintain low
impedances, comes around to get you! If you want to design a
water-cooled mic preamp, please be my guest. Just don't ask me to pay
for it!
;-)
JnyVee
walk...@thegrid.net (hank alrich) wrote:
> That's true; I never did hear him play it. It was just hanging on one of
> his livingroom walls, begging not to be played. But what does this say
> about the man if he thinks of something like _that_ as wall art? Maybe
> it's an attempt at dysenterior design.
"dysenterior design"...????
Hank! again...
(how DOES he do it folks?)
--
<Help Keep The Net Emoticon-Free>
hank alrich
Depends. 1. How many strings on the banjo? If five, it can be used as a
musical instrument. 2. Is your dart board worn out? If so, there ya go
with the velvet Elvis. Taste can be mitigated by practical application
of tasteless objects.
Next, McQ's gonna show up with pictures of himself in a canoe with the
Ovation.
Harvey Gerst
Dave Martin
: > I dunno - I think someone's gonna bring me a banjo to hang on a wall in
the
: > studio, and one room is just screaming for a Velvet Elvis...
:
: > Do those things reflect badly on my taste?
:
: Depends. 1. How many strings on the banjo? If five, it can be used as a
: musical instrument. 2. Is your dart board worn out? If so, there ya go
: with the velvet Elvis. Taste can be mitigated by practical application
: of tasteless objects.
Actually, Elvis would cover up the AC panel in one room. I just can't bring
myself to use my Ray Charles poster for that. Besides, the room is painted
with that Ralph Lauren suede paint. And it's blue...
:
: Next, McQ's gonna show up with pictures of himself in a canoe with the
: Ovation.
:
It'd blister his hands pretty quick unless the strings were removed, and
even then, once the body fills up with water it would be hard to make the
canoe go..
Bob Olhsson
Dave Martin <dave....@nashville.com> wrote:
> I think someone's gonna bring me a banjo to hang on a wall in the
>studio
You could use different sized banjos as distributed resonant traps.
--
Bob Olhsson Audio Mastery Recording Project Design and Consulting
Box 90412, Nashville TN 37209 Tracking, Mixing and Mastering
615.352.7635 FAX 615.356.2483 Mix Evaluation and Quality Control
40 years of making people sound better than they thought possible!
Dave Martin
"Bob Olhsson" <o...@hyperback.com> wrote in message
news:260820011303158606%o...@hyperback.com...
: In article <gF6i7.159082$ai2.11...@bin2.nnrp.aus1.giganews.com>,
: Dave Martin <dave....@nashville.com> wrote:
:
: > I think someone's gonna bring me a banjo to hang on a wall in the
: >studio
:
: You could use different sized banjos as distributed resonant traps.
Does that mean that I could use the bass banjo that was at NAMM? Cool!
Bob Olhsson
Dave Martin <dave....@nashville.com> wrote:
>Does that mean that I could use the bass banjo that was at NAMM?
Sure, ANY 5 string instrument...
Mark McQuilken
(snip)
Arny:
You know, I'd heard that you were just an unreasonable asshole from many
others and thought maybe I'd give you the benefit of the doubt. I was
wrong. You're clearly an intractable know-it-all. Sorry to waste my time
and yours. Thank god I don't have to buy stuff that you "design"...
Fuck-off and die,
McQ
P.S. - *PLONK* (permanent killfile)
WillStG
<< Arny:
>>You know, I'd heard that you were just an unreasonable asshole from many
others and thought maybe I'd give you the benefit of the doubt. I was wrong.
You're clearly an intractable know-it-all. Sorry to waste my time and yours. >>
Don't really know him, but Arny's seemed to be pretty much behaving
himself over here on RAP Mark. I think he just lost his bearings for a moment
and thought he was still over on RAO, where insults are normal conversation and
his kinda anal tit-for-tat post is what passes for a civil discussion.
( And isn't RAP kind of a 12 step program for RAO'er's? You know, put down
the speaker cable with arrows on it, pick up the barbeque sauce... )
Dave Martin
the usually imprecise English language as a further barrier to a discussion
that's basically mathematical in nature. I re-read his post to see what made
you reach for the ol' kill file and didn't really see anything that was
actually insulting, even if the various arguments were a little on the
sophomoric side.
Besides, you're both wrong. The real answer is 42. The implementation of
that answer is merely an engineering problem.
--
Dave (a tractable know it all) Martin
DMA, Inc.
Nashville, TN
"Mark McQuilken" <ma...@fmraudio.com> wrote in message
news:3B897E...@fmraudio.com...
:
Mixerman
have the energy to find the exact bullshit I want to comment on, so I'm
placing my
thoughts here. OK with you? Thanks.
All I have to say is this. While some schmucks (i.e. Ethan) read books and
discuss ways ad nauseum on the all-importance of reducing noise floor and
distortion to next-to-nil, I'm making a very nice living making them both
louder.
Go figure.
Mixerman
"Roger W. Norman" <rvze...@verizon.net> wrote in message
news:9mauqm$set$1...@suaar1ab.prod.compuserve.com...
David Morgan (MAMS)
"Dave Martin" <dave....@nashville.com> wrote in message...
> Besides, you're both wrong. The real answer is 42. The implementation of
> that answer is merely an engineering problem.
It's about time something in this thread made sense. We'll discuss it
when we get to the restaurant at the end of the universe.
DM
dwgriffi
<ma...@fmraudio.com> wrote:
>P.S. - *PLONK* (permanent killfile)
Welcome to the club! The first thing I ever learned about newsgroups
years ago was that rec.audio.opinion is the last place you go to get
actual information you can use about audio, unless you want to get it
from bitter guys who spend their lives trying to outdo each other and
sit alone in front of their home theater systems waiting for the
opportunity to get snippy.
When the angry Mr. Kruger drifted over here I thought, "Oh boy. He's
still not talking about audio"
DW
--
Delete Spamaway to reply
hank alrich
> Besides, you're both wrong. The real answer is 42. The implementation of
> that answer is merely an engineering problem.
Close, Dave, but no reefer; the correct answer is "four twenty"; you
missed it by a factor of ten. Back to the arithmetic.
(Maybe he's just redirecting his anger over my dysenterior design
remark.)
Mike Rivers
> I dunno - I think someone's gonna bring me a banjo to hang on a wall in the
> studio, and one room is just screaming for a Velvet Elvis...
>
> Do those things reflect badly on my taste?
If you place them properly, the reflection from the banjo head will
just about be absorbed by the velvet Elvis.
--
I'm really Mike Rivers (mri...@d-and-d.com)
Bob Olhsson
wrote:
>If you place them properly, the reflection from the banjo head will
>just about be absorbed by the velvet Elvis.
I thought you were supposed to stuff a velvet Elvis inside each banjo
in order to damp the head. The whole thing is sold as a kit at Banjo
Center although the strings are extra.
but then what do I know...
Roger W. Norman
down with the whats and whys that we don't do. I'm not going to put any
cuts on Ethan. He's an excellent composer and realizes his music the way he
feels comfortable. I proud to have his work in my collection and his
ability to take an orchestration and show a day in the life of his cat fully
fits with my cat. If he can do that he speaks well in his music. Make no
mistake, I like Ethan and have had a cyber relationship with him for maybe 6
or 7 years now. But we don't always see eye to eye, and that's probably my
ignorance showing through, not something wrong with him.
BUT, on this he and I have never seen eye to eye, and I'd still invite him
to any bbq I ever threw if he was in the area. Shit, if he was in the area,
I'd throw a bbq just to have him over!
Then again, I'd do it for most any of you guys. This group has become my
second home, although I've learned that I don't have too many reasons to
post here as much as I used to because I'm just not up to snuff. Another
twenty years or so and just maybe.... (That's why I come bug you on RO!
<g>)
--
Roger W. Norman
www.SirMusicStudio.com
Ro...@SirMusicStudio.com
301-585-4681
"guys, it takes a lifetime to just get just a BIT closer..."
George Massenburg
"Mixerman" <Mixe...@verizon.net> wrote in message
news:JCji7.2619$BH1.1...@paloalto-snr1.gtei.net...
Roger W. Norman
how to make people think. So the point of this is that while I was reading
through the tread, I caught that Tom Cram had his website listed, which I
don't believe he's done here, so I listened. This is the type of music I
was hoping you'd see out of the boyz I sent you mp3s of. Not the type of
music you necessarily do, but I think you'd appreciate it. Killer
recordings, particular for mp3s. I'm buying the CD.
--
Roger W. Norman
www.SirMusicStudio.com
Ro...@SirMusicStudio.com
301-585-4681
"guys, it takes a lifetime to just get just a BIT closer..."
George Massenburg
"Mixerman" <Mixe...@verizon.net> wrote in message
news:JCji7.2619$BH1.1...@paloalto-snr1.gtei.net...
Dave Martin
"Mike Rivers" <mri...@d-and-d.com> wrote in message
news:znr998880662k@trad...
:
: In article <gF6i7.159082$ai2.11...@bin2.nnrp.aus1.giganews.com>
dave....@nashville.com writes:
:
: > I dunno - I think someone's gonna bring me a banjo to hang on a wall in
the
: > studio, and one room is just screaming for a Velvet Elvis...
: >
: > Do those things reflect badly on my taste?
:
: If you place them properly, the reflection from the banjo head will
: just about be absorbed by the velvet Elvis.
I'm afraid that the technology police will come if I hang a banjo in the Pro
Tools suite, so I may have to look at other alternatives, like the dogs
playing poker...
Dave Martin
: In article <znr998880662k@trad>, Mike Rivers <mri...@d-and-d.com>
: wrote:
:
: >If you place them properly, the reflection from the banjo head will
: >just about be absorbed by the velvet Elvis.
:
: I thought you were supposed to stuff a velvet Elvis inside each banjo
: in order to damp the head. The whole thing is sold as a kit at Banjo
: Center although the strings are extra.
:
: but then what do I know...
You're close - I think you're supposed to stuff an accordion inside each
banjo.
Roger W. Norman
off and die". Must have been a bad day.
--
Roger W. Norman
www.SirMusicStudio.com
Ro...@SirMusicStudio.com
301-585-4681
"guys, it takes a lifetime to just get just a BIT closer..."
George Massenburg
"Mark McQuilken" <ma...@fmraudio.com> wrote in message
news:3B897E...@fmraudio.com...
Roger W. Norman
--
Roger W. Norman
www.SirMusicStudio.com
Ro...@SirMusicStudio.com
301-585-4681
"guys, it takes a lifetime to just get just a BIT closer..."
George Massenburg
"Dave Martin" <dave....@nashville.com> wrote in message
news:kuii7.82084$Lw3.5...@news2.aus1.giganews.com...
Dave Martin
news:233CB3B18B571C0C.343E22B1...@lp.airnews.net...
:
: "Dave Martin" <dave....@nashville.com> wrote in message...
I'm not sure, but I think that's a catfish house in Cheatham County.
--
Dave Martin
DMA, Inc.
Nashville, TN
Roger W. Norman
make the banjo head bulge.
--
Roger W. Norman
www.SirMusicStudio.com
Ro...@SirMusicStudio.com
301-585-4681
"guys, it takes a lifetime to just get just a BIT closer..."
George Massenburg
"Bob Olhsson" <o...@hyperback.com> wrote in message
news:270820010910375846%o...@hyperback.com...
Roger W. Norman
banjo into the dumpster and it lands on the accordian.
If you stuffed an accordian inside of a banjo wouldn't you get cajun music?
--
Roger W. Norman
www.SirMusicStudio.com
Ro...@SirMusicStudio.com
301-585-4681
"guys, it takes a lifetime to just get just a BIT closer..."
George Massenburg
"Dave Martin" <dave....@nashville.com> wrote in message
news:Tgti7.122975$NK1.10...@bin3.nnrp.aus1.giganews.com...
Roger W. Norman
know if he's typing with one hand or two, now do we? <g>
--
Roger W. Norman
www.SirMusicStudio.com
Ro...@SirMusicStudio.com
301-585-4681
"guys, it takes a lifetime to just get just a BIT closer..."
George Massenburg
"JnyVee" <moc....@ybmurbrevlis.com> wrote in message
news:260820011215503465%moc....@ybmurbrevlis.com...
Mike Rivers
> You're close - I think you're supposed to stuff an accordion inside each
> banjo.
No, no - you stuff the banjo INSIDE the accordion. You might have to
break the neck in two, but just be sure that at least one piece is
long enough so that it keeps the bellows fully extended.
Scott Dorsey
>In article <Tgti7.122975$NK1.10...@bin3.nnrp.aus1.giganews.com> dave....@nashville.com writes:
>
>> You're close - I think you're supposed to stuff an accordion inside each
>> banjo.
>
>No, no - you stuff the banjo INSIDE the accordion. You might have to
>break the neck in two, but just be sure that at least one piece is
>long enough so that it keeps the bellows fully extended.
I don't get it. If you do this, there won't be room for the bagpipes.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."
Dave Martin
--
Dave Martin
DMA, Inc.
Nashville, TN
"Roger W. Norman" <rvze...@verizon.net> wrote in message
news:9mdq3g$i26$1...@suaar1ab.prod.compuserve.com...
: No, it's 4'2. Get it right Dave.
: > :
: >
: >
:
:
:
hank alrich
> I'm afraid that the technology police will come if I hang a banjo in the Pro
> Tools suite, so I may have to look at other alternatives, like the dogs
> playing poker...
Two words: "Bela Fleck". That'll stall the tech police long enough for
you to make an escape.
Ben Bradley
>
>In article <Tgti7.122975$NK1.10...@bin3.nnrp.aus1.giganews.com> dave....@nashville.com writes:
>
>> You're close - I think you're supposed to stuff an accordion inside each
>> banjo.
>
>No, no - you stuff the banjo INSIDE the accordion. You might have to
>break the neck in two, but just be sure that at least one piece is
>long enough so that it keeps the bellows fully extended.
If the bellows gets punctured, is that a problem? Or is that a
solution?
Regardless, I wouldn't rely solely on a puncture, as the
accordionist might find some duct tape.
>--
>I'm really Mike Rivers (mri...@d-and-d.com)