[Steve Byers]

In your experience, assuming you use quality converters, how many
conversions (from analog to digital and back again) can typically be
performed before discernible quality loss is incurred?

[Laurence Payne[_2_]]

On Wed, 02 Jul 2008 14:55:45 GMT, "Steve Byers"
wrote:

In your experience, assuming you use quality converters, how many
conversions (from analog to digital and back again) can typically be
performed before discernible quality loss is incurred?

Depends how discerning you are, and on the material. I'd be getting
very worried if I had to go from analogue to wav more than 3 or 4
times, even on rock music.

[Scott Dorsey]

Steve Byers wrote:
In your experience, assuming you use quality converters, how many
conversions (from analog to digital and back again) can typically be
performed before discernible quality loss is incurred?

I could do about thirty passes on the Prism AD-124 before it became
obvious. The Prism is by no means state of the art, though.

It's a fun test to make on your own hardware. Linearity issues get
very exaggerated.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Don Pearce]

Scott Dorsey wrote:
Steve Byers wrote:
In your experience, assuming you use quality converters, how many
conversions (from analog to digital and back again) can typically be
performed before discernible quality loss is incurred?

I could do about thirty passes on the Prism AD-124 before it became
obvious. The Prism is by no means state of the art, though.

It's a fun test to make on your own hardware. Linearity issues get
very exaggerated.

As do deviations from a ruler-flat frequency response. I've tried this
before with an Audiophile 2496, and 10 passes were still perfectly
acceptable.

d

[Arny Krueger]

"Steve Byers" wrote in message


In your experience, assuming you use quality converters,
how many conversions (from analog to digital and back
again) can typically be performed before discernible
quality loss is incurred?

Depends on lots of things, the source material and the monitoring
environment being at or near the top of the considerations.

The two major audible converter deficiencies that still affect modern
converters relate to frequency response and dynamic range.

For example, take a converter pair that are off by 0.1 dB at 5 KHz, in an
octave-wide band. This is actually a pretty high number for a good set of
pro-grade converters. 10 passes and now you have a 1.0 dB variation and that
could be audible. 5 passes is only 0.5 dB, and that will probably slip by.

For another example, take a converter pair that have 100 dB dynamic range -
good quality converters will meet or beat this. Every doubling of passes
will cost you 3 dB. It will take a lot of passes before the noise floor
gets within 10 dB of a typical recording with 70 dB or poorer dynamic range.

If the converters have signal-related spurious responses, then they are
coherent and have fixed phase relationships and will add linearly, just like
the frequency response variations mentioned above.

As a rule of thumb, 5 passes might produce slightly noticeable differences,
but maybe not. If you have really top-grade equipment then 10-20 passes
should still ride free.

If you want to play with this problem - start out trying looping-back the
audio interface that came with the system board in your PC. You might start
getting some action after only 3-5 passes, but probably more than 1.

As far as procedures go, make up a test selection from one of your best
recordings, and throw a level-set tone into the set. Then use your level-set
tone to make sure that you are comparing apples to apples, and use one of
the DBT comparators that you can download from the net such as this one:

http://www.kikeg.arrakis.es/winabx/

[David Morgan \(MAMS\)]

"Scott Dorsey" wrote in message ...
Steve Byers wrote:
In your experience, assuming you use quality converters, how many
conversions (from analog to digital and back again) can typically be
performed before discernible quality loss is incurred?

I could do about thirty passes on the Prism AD-124 before it became
obvious. The Prism is by no means state of the art, though.

It's a fun test to make on your own hardware. Linearity issues get
very exaggerated.


OK... I have a serious question for you. I'm about to *consider* repairing
some ProTools mixes by passing the mix out a couple of channels and
adding a bit more of a fill instrument or two, by doing this in the analogue
world and recording the new mix back to PT. Do you think I'd be damaging
the material more in this way, than I would by trying to re-set a PT mix which
was done completely in the analogue world and NOT in the box? (I'm on
Digi 192s clocked with an Apogee Rosetta, PT 7.2.1, HD2).

There's some magic in the mix... it's just a little shy of some parts, and
re-creating the mix seems like a monumental task since it wasn't done
in the box to begin with.

Thanks,

DM

[Scott Dorsey]

David Morgan \(MAMS\) /Odm wrote:

OK... I have a serious question for you. I'm about to *consider* repairing
some ProTools mixes by passing the mix out a couple of channels and
adding a bit more of a fill instrument or two, by doing this in the analogue
world and recording the new mix back to PT. Do you think I'd be damaging
the material more in this way, than I would by trying to re-set a PT mix which
was done completely in the analogue world and NOT in the box? (I'm on
Digi 192s clocked with an Apogee Rosetta, PT 7.2.1, HD2).

Why not just import the full mix that you have into two channels of PT,
then lay down the tracks you want and mix them together? That avoids
going through any additional converters, it avoids the analogue summing,
and it allows you to cut and paste the new material if you decide you want to.
--scott

--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Tobiah]

OK... I have a serious question for you. I'm about to *consider* repairing
some ProTools mixes by passing the mix out a couple of channels and
adding a bit more of a fill instrument or two, by doing this in the analogue
world and recording the new mix back to PT.

Are you saying that you want to send the old mix out and loop record
it back again with new instruments added? Why wouldn't you just record
the new instruments on new tracks?


** Posted from http://www.teranews.com **

[Earl Kiosterud]

"Steve Byers" wrote in message
...
In your experience, assuming you use quality converters, how many conversions (from analog
to digital and back again) can typically be performed before discernible quality loss is
incurred?

If the program material and level isn't exactly the same, and it won't be, you can expect a
nominal 3 dB loss in S/N on each pass in the recording side, through the ADC, even if the
converters are perfect. In the playback side, if there are any level-adjusting items, like
the Windows virtual device driver, which adjusts the playback level on the digital side of
the DAC (in response to moving the on-screen fader), then there goes another 3 dB nominally,
on each pass, unless the playback attenuation is exactly 0 (or 1/2 or 1/4, etc.). That's 6
dB total per pass, even before any imperfections in the ADC or DAC or other stuff. If the
audio software has a playback level control (not counting one that directly controls the
levels on the analog playback side of the soundcard -- most don't -- they do it in digital
before it's sent to the soundcard), there's another 3 dB, again unless the attenuation is 0
or 1/2, etc.

If you're talking about one or two passes, don't worry about it at all -- you'll never hear
the difference.
--
Earl

[Laurence Payne[_2_]]

On Wed, 02 Jul 2008 17:49:38 GMT, "David Morgan \(MAMS\)"
/Odm wrote:

OK... I have a serious question for you. I'm about to *consider* repairing
some ProTools mixes by passing the mix out a couple of channels and
adding a bit more of a fill instrument or two, by doing this in the analogue
world and recording the new mix back to PT. Do you think I'd be damaging
the material more in this way, than I would by trying to re-set a PT mix which
was done completely in the analogue world and NOT in the box? (I'm on
Digi 192s clocked with an Apogee Rosetta, PT 7.2.1, HD2).

Why do you want to mix on-the-fly? Leave the PT stuff in PT, add a
couple more tracks, adjust them to suit the mix. Exactly the same
result, but all the time ithe world to get the mix right.

[Don Pearce]

Steve Byers wrote:
In your experience, assuming you use quality converters, how many
conversions (from analog to digital and back again) can typically be
performed before discernible quality loss is incurred?



You can make your own comparison here. A few seconds of first
generation, followed by the same clip at tenth generation - done with
reasonable care. This with a standard consumer-grade sound card.

http://81.174.169.10/odds/gen10.wav

What do you think - discernible or not?

d

[Steve Byers]

Thank you all for your helpful replies.

[Mike Rivers]

Steve Byers wrote:
In your experience, assuming you use quality converters, how many
conversions (from analog to digital and back again) can typically be
performed before discernible quality loss is incurred?

Would it be unfair to say "I don't care about quality loss."? If I
needed to make a conversion to improve the product, then I'd do it. If
there was more loss than gain, then I'd un-do it and try something else.
If it was better after I did whatever I did, then there is no quality
loss, there's quality gained.

I wouldn't pass a signal in and out and in and out and in and out of a
converter for no good reason. But if I had a good reason, I wouldn't
worry about it. Heck, I copied analog tape to analog tape to analog tape
many times and lived to tell the tale. But I know some people get so
hung up about once-it's-digital-it-must-stay-digital-or-it's-ruined that
they worry about things like this.

Let me reply for you: "I'm not worried, I just wanted to know for my
own information."

Try it.



--
If you e-mail me and it bounces, use your secret decoder ring and reach
me he
double-m-eleven-double-zero at yahoo -- I'm really Mike Rivers
)

[david correia]

In article ,
(Scott Dorsey) wrote:

Steve Byers wrote:
In your experience, assuming you use quality converters, how many
conversions (from analog to digital and back again) can typically be
performed before discernible quality loss is incurred?

I could do about thirty passes on the Prism AD-124 before it became
obvious. The Prism is by no means state of the art, though.

It's a fun test to make on your own hardware. Linearity issues get
very exaggerated.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."



Sometime during the 80's I bounced one song 4 times between a couple
Sony DAT recorders (analog in and out of course) and was very surprised
to find I couldn't hear any difference.




David Correia
www.Celebrationsound.com

[David Morgan \(MAMS\)]

"Scott Dorsey" wrote in message ...
David Morgan \(MAMS\) /Odm wrote:

OK... I have a serious question for you. I'm about to *consider* repairing
some ProTools mixes by passing the mix out a couple of channels and
adding a bit more of a fill instrument or two, by doing this in the analogue
world and recording the new mix back to PT. Do you think I'd be damaging
the material more in this way, than I would by trying to re-set a PT mix which
was done completely in the analogue world and NOT in the box? (I'm on
Digi 192s clocked with an Apogee Rosetta, PT 7.2.1, HD2).

Why not just import the full mix that you have into two channels of PT,
then lay down the tracks you want and mix them together? That avoids
going through any additional converters, it avoids the analogue summing,
and it allows you to cut and paste the new material if you decide you want to.
--scott

I mixed in the analogue world, but mixed back to PT... so the tracks *are*
in ProTools.... meaning this would be pass number 3 and 4 through the A/D.

*Everything* is in PT, with the exception of the hardware world I mixed through,
including my needed volume automation on the tracks I think need to be boosted
a little... and most certainly the mixes, but probably a couple of milliseconds or so
behind the tracks; I'll figure that one out later. I have no issues with the analogue
'summing' of three or four tracks on the MCI.

If you can do 30 passes on a Prism, are you insinuating that I shouldn't
do four on Digi 192's through an Apogee Rosetta clock?

Admittedly, I'm looking for an easy way out.

Tell you what... If you think there's enough steel guitar in this unmastered
mix, I'll blow off the whole idea. I think it needs more.
http://www.m-a-m-s.com/MP3/BTZ/Heaven_In_Your_Arms.mp3



Cheers,

DM

[David Morgan \(MAMS\)]

"Tobiah" wrote in message news

OK... I have a serious question for you. I'm about to *consider* repairing
some ProTools mixes by passing the mix out a couple of channels and
adding a bit more of a fill instrument or two, by doing this in the analogue
world and recording the new mix back to PT.

Are you saying that you want to send the old mix out and loop record
it back again with new instruments added? Why wouldn't you just record
the new instruments on new tracks?

The instruments are already there. The mix, done in the analogue
world through real hardware, is there too... and supposedly finished.
Speaking for myself, Elvis left the building without getting enough
of a particular instrument in a rather complicated hardware-world
mix.

I was thinking the quickest way to satisfy my needs would be to
assign the stereo mix to a couple of tracks, solo the mix tracks
and the instrument(s) I want to increase slightly in volume. Then,
mix the three or four total tracks in the analogue world, back into a
couple more PT tracks in the same session folder.

DM

[David Morgan \(MAMS\)]

"Laurence Payne" wrote in message news
On Wed, 02 Jul 2008 17:49:38 GMT, "David Morgan \(MAMS\)"
/Odm wrote:

OK... I have a serious question for you. I'm about to *consider* repairing
some ProTools mixes by passing the mix out a couple of channels and
adding a bit more of a fill instrument or two, by doing this in the analogue
world and recording the new mix back to PT. Do you think I'd be damaging
the material more in this way, than I would by trying to re-set a PT mix which
was done completely in the analogue world and NOT in the box? (I'm on
Digi 192s clocked with an Apogee Rosetta, PT 7.2.1, HD2).

Why do you want to mix on-the-fly? Leave the PT stuff in PT, add a
couple more tracks, adjust them to suit the mix. Exactly the same
result, but all the time ithe world to get the mix right.

I'm not sure if I follow you here.

I'm using Pro Tools as a tape machine with some volume automation
and nothing more. I'm bouncing my mixes back to a couple of tracks.
I need more volume on one instrument from a mix done *outside* PT,
but recorded back to PT, using tracks from *inside* PT. Everything
else in the mix path but the tracks, was pure analogue hardware.


DM

[Don Pearce]

David Morgan (MAMS) wrote:
"Scott Dorsey" wrote in message ...
David Morgan \(MAMS\) /Odm wrote:
OK... I have a serious question for you. I'm about to *consider* repairing
some ProTools mixes by passing the mix out a couple of channels and
adding a bit more of a fill instrument or two, by doing this in the analogue
world and recording the new mix back to PT. Do you think I'd be damaging
the material more in this way, than I would by trying to re-set a PT mix which
was done completely in the analogue world and NOT in the box? (I'm on
Digi 192s clocked with an Apogee Rosetta, PT 7.2.1, HD2).

Why not just import the full mix that you have into two channels of PT,
then lay down the tracks you want and mix them together? That avoids
going through any additional converters, it avoids the analogue summing,
and it allows you to cut and paste the new material if you decide you want to.
--scott

I mixed in the analogue world, but mixed back to PT... so the tracks *are*
in ProTools.... meaning this would be pass number 3 and 4 through the A/D.

*Everything* is in PT, with the exception of the hardware world I mixed through,
including my needed volume automation on the tracks I think need to be boosted
a little... and most certainly the mixes, but probably a couple of milliseconds or so
behind the tracks; I'll figure that one out later. I have no issues with the analogue
'summing' of three or four tracks on the MCI.

If you can do 30 passes on a Prism, are you insinuating that I shouldn't
do four on Digi 192's through an Apogee Rosetta clock?

Admittedly, I'm looking for an easy way out.

Tell you what... If you think there's enough steel guitar in this unmastered
mix, I'll blow off the whole idea. I think it needs more.
http://www.m-a-m-s.com/MP3/BTZ/Heaven_In_Your_Arms.mp3



Cheers,

DM


Plenty enough steel there. If I have any problem with that mix it is
that the drums appear to occupy a different acoustic space to the rest
of the mix. And the arrangement is all just a bit relentless - it needs
some contrast with everyone backing off a bit, some kind of middle eight.

d

[Arny Krueger]

"Earl Kiosterud" wrote in message
news:35Qak.365$Ae3.101@trnddc05
"Steve Byers" wrote in message
...

In your experience, assuming you use quality converters,
how many conversions (from analog to digital and back
again) can typically be performed before discernible
quality loss is incurred?

If the program material and level isn't exactly the same,
and it won't be, you can expect a nominal 3 dB loss in
S/N on each pass in the recording side, through the ADC,
even if the converters are perfect.

Not quite how it works.

The 3 dB increase in noise only takes place if the source material has the
same dynamic range as the converters.

Because of this, each 3 dB increase of overall noise level requires a
doubling of the number of passes.

IOW, if you have source material with a typical 70 dB dynamic range, and
pass it through some converters with a typical 100 dB dynamic range, the
resulting file will have dynamic range that is so close to 70 dB that it
will be difficult or impossible to measure the difference.

It is true that if you have source material with 100 dB dynamic range, and
pass it through converters with 100 dB dynamic range, then the resulting
file will have only 97 dB dynamic range.

However, if you pass the file with 97 dB dynamic range thrhough the same
converters again, the increase in noise will be somewhat less than 3 dB.
The next time the increase will be even less than 3 dB. After a while, you
will have a file with 90 dB dynamic range, and the increase in noise level
for each pass will only be a few tenths of a dB.

[Scott Dorsey]

David Morgan \(MAMS\) /Odm wrote:
I mixed in the analogue world, but mixed back to PT... so the tracks *are*
in ProTools.... meaning this would be pass number 3 and 4 through the A/D.

*Everything* is in PT, with the exception of the hardware world I mixed through,
including my needed volume automation on the tracks I think need to be boosted
a little... and most certainly the mixes, but probably a couple of milliseconds or so
behind the tracks; I'll figure that one out later. I have no issues with the analogue
'summing' of three or four tracks on the MCI.

So take the tracks that have the existing mixes, and add the new tracks
to them, as if they were a stem.

As long as you don't have any effects that have been applied on the 2-buss,
this should be close to the same as summing analogue.

If you can do 30 passes on a Prism, are you insinuating that I shouldn't
do four on Digi 192's through an Apogee Rosetta clock?

Damned if I know, I never tried it. I _did_ try it with the Prism.

Admittedly, I'm looking for an easy way out.

Tell you what... If you think there's enough steel guitar in this unmastered
mix, I'll blow off the whole idea. I think it needs more.
http://www.m-a-m-s.com/MP3/BTZ/Heaven_In_Your_Arms.mp3

I can't play it right now but I will give it a listen on Tuesday when I am
back at a modern computer.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

[Predrag Trpkov]

"David Morgan (MAMS)" /Odm wrote in message
newsZ_ak.505$bn3.238@trnddc07...

"Scott Dorsey" wrote in message
...
David Morgan \(MAMS\) /Odm wrote:

OK... I have a serious question for you. I'm about to *consider*
repairing
some ProTools mixes by passing the mix out a couple of channels and
adding a bit more of a fill instrument or two, by doing this in the
analogue
world and recording the new mix back to PT. Do you think I'd be
damaging
the material more in this way, than I would by trying to re-set a PT
mix which
was done completely in the analogue world and NOT in the box? (I'm on
Digi 192s clocked with an Apogee Rosetta, PT 7.2.1, HD2).

Why not just import the full mix that you have into two channels of PT,
then lay down the tracks you want and mix them together? That avoids
going through any additional converters, it avoids the analogue summing,
and it allows you to cut and paste the new material if you decide you
want to.
--scott

I mixed in the analogue world, but mixed back to PT... so the tracks *are*
in ProTools.... meaning this would be pass number 3 and 4 through the A/D.

*Everything* is in PT, with the exception of the hardware world I mixed
through,
including my needed volume automation on the tracks I think need to be
boosted
a little... and most certainly the mixes, but probably a couple of
milliseconds or so
behind the tracks; I'll figure that one out later. I have no issues with
the analogue
'summing' of three or four tracks on the MCI.

If you can do 30 passes on a Prism, are you insinuating that I shouldn't
do four on Digi 192's through an Apogee Rosetta clock?

Admittedly, I'm looking for an easy way out.

Tell you what... If you think there's enough steel guitar in this
unmastered
mix, I'll blow off the whole idea. I think it needs more.
http://www.m-a-m-s.com/MP3/BTZ/Heaven_In_Your_Arms.mp3



This mix sounds good to me as it is, but I can imagine it benefitting
somewhat from making the steel guitar as prominent as the acoustic guitar
fill-ins on the opposite side.

You can do it both in and out of the box. Just align the tracks visually as
well as you can and it should be a breeze.

It's not the easiest kind of arrangement to work on, so many busy guitars. I
like what you've done. How long did it take you to mix it? What MCI console?

Predrag

[David Morgan \(MAMS\)]

"Scott Dorsey" wrote in message...

David Morgan \(MAMS\) /Odm wrote:

Admittedly, I'm looking for an easy way out.

Tell you what... If you think there's enough steel guitar in this unmastered
mix, I'll blow off the whole idea. I think it needs more.
http://www.m-a-m-s.com/MP3/BTZ/Heaven_In_Your_Arms.mp3

I can't play it right now but I will give it a listen on Tuesday when I am
back at a modern computer.

I'll be back home on Sunday night, and the tentative adte for trying
this little trick will be Wednesday afternoon.

Cheers,

DM

[Ethan Winer]

how many conversions (from analog to digital and back again) can typically
be performed before discernible quality loss is incurred?

I actually tested that with a $25 SoundBlaster card. I got a lot more
degradation than others report with "good" converters, but if you'd like to
hear the result here are links to the original Wave, after one pass, ten
passes, and twenty passes. Each file is about 2 MB:

http://www.ethanwiner.com/misc-conte...x_original.wav
www.ethanwiner.com/misc-content/sb20x_pass1.wav
www.ethanwiner.com/misc-content/sb20x_pass10.wav
www.ethanwiner.com/misc-content/sb20x_pass20.wav

--Ethan

[David Morgan \(MAMS\)]

"Predrag Trpkov" wrote in message:

"David Morgan (MAMS)" /Odm wrote ...

Tell you what... If you think there's enough steel guitar in this
unmastered mix, I'll blow off the whole idea. I think it needs more.
http://www.m-a-m-s.com/MP3/BTZ/Heaven_In_Your_Arms.mp3

This mix sounds good to me as it is, but I can imagine it benefitting
somewhat from making the steel guitar as prominent as the acoustic
guitar fill-ins on the opposite side.

I agree completely. The singer had a favorite steel 'lick' in the last
chorus and had me turn it up. That ended up sounding awkward to
me, but it's a reference point for where the steel should have been
throughout the entire song. The whole problem here, was that I did
not have my little 6" Tannoys with me for the mix, so I mixed on the
big TAD's and headphones... and they apparently both lied to me
about the presence of the steel frequencies.

You can do it both in and out of the box. Just align the tracks visually
as well as you can and it should be a breeze.

That's not so easy. In millisecond increments, 2 is not enough,
but three is too many. :-(

I've already busted my ass trying to do this inside the box. The old
"sum of the parts" principle (along with my unfamiliarity with the inner
workings of PT), has wasted too much time already. The mix, while
pretty dense, requires that I severely turn down the steel guitar track
level in PT to make it fit with the mix. If I do this, I'm losing all of the
automation that I've written to that track. I'm sure there's a way to turn
down the whole track level and maintain the automation, but I can't find
it. Back to the 'sum of the parts' thing... this is one of those cases
where a dB or so (or even less) of extra presence of the raw
instrument track, combined with the overall mix, could be too much.
But I had rather chance that, than take the 5 hours or so needed to
re-establish the analogue mix and maybe miss some of the magic.

It's not the easiest kind of arrangement to work on, so many busy
guitars. I like what you've done. How long did it take you to mix it?

There are three tracks of electric (one stacked rhythm and a solo), four
tracks of acoustic rhythm (with one stereo take being 2 of those) , and
an acoustic solo; most all of if going through a multitude of outboard
processing and FX. That's an 8 hour mix, which includes the spot
editing and writing the track automation for about 22 tracks, IIRC.

What MCI console?

It's a 536... and if I do say so myself (as do the Blevins people),
probably the most pristine 536 left in the US. There's nothing on
it that doesn't work, but it's not wired for 'quad'. :-)

I'm pretty sure I'm going to take a run at this in the analogue world.
The guys at the studio are screaming "quality loss".... but I'm in the
same mindset as Mike Rivers' post of last night to this thread.


Cheers,

DM



--
David Morgan (MAMS)
Morgan Audio Media Service
http://www.m-a-m-s DOT com
Dallas, Texas (214) 662-9901
_____________________________
http://www.januarysound.com

[philicorda[_6_]]

On Thu, 03 Jul 2008 12:58:12 -0400, Ethan Winer wrote:

how many conversions (from analog to digital and back again) can
typically be performed before discernible quality loss is incurred?

I actually tested that with a $25 SoundBlaster card. I got a lot more
degradation than others report with "good" converters, but if you'd like
to hear the result here are links to the original Wave, after one pass,
ten passes, and twenty passes. Each file is about 2 MB:

http://www.ethanwiner.com/misc-conte...x_original.wav
www.ethanwiner.com/misc-content/sb20x_pass1.wav
www.ethanwiner.com/misc-content/sb20x_pass10.wav
www.ethanwiner.com/misc-content/sb20x_pass20.wav

Wow. It's quite astonishing how it sounds after 20 conversions, for a $25
card. Perhaps a little blurry and lacking in punch by comparison, but
certainly not awful.
Was it a SoundBlaster Live?

It makes me think that converters are unjustly blamed for bad sound
quality far too often.


--Ethan

[Federico]

My idea is that the problem about "bad sound" lies in the transformerless
modern audio chain...
F.

It makes me think that converters are unjustly blamed for bad sound
quality far too often.

[Romeo Rondeau[_4_]]

David Morgan (MAMS) wrote:
"Predrag Trpkov" wrote in message:

"David Morgan (MAMS)" /Odm wrote ...

Tell you what... If you think there's enough steel guitar in this
unmastered mix, I'll blow off the whole idea. I think it needs more.
http://www.m-a-m-s.com/MP3/BTZ/Heaven_In_Your_Arms.mp3

This mix sounds good to me as it is, but I can imagine it benefitting
somewhat from making the steel guitar as prominent as the acoustic
guitar fill-ins on the opposite side.

I agree completely. The singer had a favorite steel 'lick' in the last
chorus and had me turn it up. That ended up sounding awkward to
me, but it's a reference point for where the steel should have been
throughout the entire song. The whole problem here, was that I did
not have my little 6" Tannoys with me for the mix, so I mixed on the
big TAD's and headphones... and they apparently both lied to me
about the presence of the steel frequencies.

You can do it both in and out of the box. Just align the tracks visually
as well as you can and it should be a breeze.

That's not so easy. In millisecond increments, 2 is not enough,
but three is too many. :-(

I've already busted my ass trying to do this inside the box. The old
"sum of the parts" principle (along with my unfamiliarity with the inner
workings of PT), has wasted too much time already. The mix, while
pretty dense, requires that I severely turn down the steel guitar track
level in PT to make it fit with the mix. If I do this, I'm losing all of the
automation that I've written to that track. I'm sure there's a way to turn
down the whole track level and maintain the automation, but I can't find
it. Back to the 'sum of the parts' thing... this is one of those cases
where a dB or so (or even less) of extra presence of the raw
instrument track, combined with the overall mix, could be too much.
But I had rather chance that, than take the 5 hours or so needed to
re-establish the analogue mix and maybe miss some of the magic.

Hey Dave! Take the steel track with the automation, assign it to it's
own buss, create a blank track using that same buss as it's input, put
the new track into record, thus bouncing the old track's automation to
disk. Mute the old track and trim with the new tracks fader.

[Mike Rivers]

Romeo Rondeau wrote:

Hey Dave! Take the steel track with the automation, assign it to it's
own buss, create a blank track using that same buss as it's input, put
the new track into record, thus bouncing the old track's automation to
disk. Mute the old track and trim with the new tracks fader.

That's just like doing it with a real mixer, only it sounds a lot more
complicated.


--
If you e-mail me and it bounces, use your secret decoder ring and reach
me he
double-m-eleven-double-zero at yahoo -- I'm really Mike Rivers
)

[Earl Kiosterud]

"Arny Krueger" wrote in message
. ..
"Earl Kiosterud" wrote in message
news:35Qak.365$Ae3.101@trnddc05
"Steve Byers" wrote in message
...

In your experience, assuming you use quality converters,
how many conversions (from analog to digital and back
again) can typically be performed before discernible
quality loss is incurred?

If the program material and level isn't exactly the same,
and it won't be, you can expect a nominal 3 dB loss in
S/N on each pass in the recording side, through the ADC,
even if the converters are perfect.

Not quite how it works.

The 3 dB increase in noise only takes place if the source material has the same dynamic
range as the converters.

Because of this, each 3 dB increase of overall noise level requires a doubling of the
number of passes.

IOW, if you have source material with a typical 70 dB dynamic range, and pass it through
some converters with a typical 100 dB dynamic range, the resulting file will have dynamic
range that is so close to 70 dB that it will be difficult or impossible to measure the
difference.

It is true that if you have source material with 100 dB dynamic range, and pass it through
converters with 100 dB dynamic range, then the resulting file will have only 97 dB dynamic
range.

However, if you pass the file with 97 dB dynamic range thrhough the same converters again,
the increase in noise will be somewhat less than 3 dB. The next time the increase will be
even less than 3 dB. After a while, you will have a file with 90 dB dynamic range, and
the increase in noise level for each pass will only be a few tenths of a dB.


You're right. I shure enough should've proofed that one before I hit send! Sorry folks.
The first ADC sets a noise level. The second pass worsens it by 3 dB. It would take two
more to worsen it by another 3 dB.

The point I wanted to make was that every multiplier (digitally implemented volume control)
in the signal path has the same effect as another pass through a perfect ADC, such as in the
virtual device driver, or a playback level control in the audio software. There are more
effective "passes" than it first seems.
--
Earl

[Romeo Rondeau[_4_]]

Mike Rivers wrote:
Romeo Rondeau wrote:

Hey Dave! Take the steel track with the automation, assign it to it's
own buss, create a blank track using that same buss as it's input, put
the new track into record, thus bouncing the old track's automation to
disk. Mute the old track and trim with the new tracks fader.

That's just like doing it with a real mixer, only it sounds a lot more
complicated.

Yeah, except you don't have another generation in electronics. It only
sounds complicated :-)

[Arny Krueger]

"Earl Kiosterud" wrote in message
news:Qdfbk.639$bn3.109@trnddc07

The point I wanted to make was that every multiplier
(digitally implemented volume control) in the signal path
has the same effect as another pass through a perfect
ADC, such as in the virtual device driver, or a playback
level control in the audio software. There are more
effective "passes" than it first seems.

Right. The good news is that these days a lot of processing is done in DSPs.
There are two benefits of this:

(1) Hopefully, authors of code for DSP-based audio processors will agregate
their processing, and do helpful thing like figure out total attenuation
separately, and actually apply it only once.

(2) Most modern DSPs for professional audio seem to have arithmetic
precision coming out of their ears, and processing precision concerns are
overkilled.

[Arny Krueger]

"Federico" wrote in message


It makes me think that converters are unjustly blamed
for bad sound quality far too often.

My idea is that the problem about "bad sound" lies in the
transformerless modern audio chain...


My thought is that the nut turning the knobs is most often the problem. ;-)

Umm, guilty as charged far more than I would like to think.

[Arny Krueger]

"Ethan Winer" ethanw at ethanwiner dot com wrote in
message
how many conversions (from analog to digital and back
again) can typically be performed before discernible
quality loss is incurred?

I actually tested that with a $25 SoundBlaster card. I
got a lot more degradation than others report with "good"
converters, but if you'd like to hear the result here are
links to the original Wave, after one pass, ten passes,
and twenty passes. Each file is about 2 MB:
http://www.ethanwiner.com/misc-conte...x_original.wav
www.ethanwiner.com/misc-content/sb20x_pass1.wav
www.ethanwiner.com/misc-content/sb20x_pass10.wav
www.ethanwiner.com/misc-content/sb20x_pass20.wav

Just for grins I made a composite stereo track of the left channel of the
original on the left, and the left channel of the 20th pass on the right.

There are some expected things like a 4 dB loss at 20 Hz, and an 11 dB loss
at 20 kHz.

This tells you what sort of fairly crude (by modern standards) converter is
in play. It's about 0.5 dB down at 20 KHz, while the good ones are less than
0.1 dB down.

You say $25? My, times have changed! Back around 1996 the then current SB64
cost nearly $200 and was very audible in one pass stereo, full duplex.

Not expected was an approximate 0.3% frequency shift. IOW an artifact of
the original recording was shifted from 1003 Hz to 1000 Hz, and this was
consistent across the whole frequency spectrum.

Of course ABXing the two was a slam dunk.

Not so with the pass1 file. I suspect the fun begins around pass 5. It
seems like a really good converter might do 10 passes and still pass an ABX
test.

[Roger Norman]

OK... I have a serious question for you. I'm about to *consider*
repairing
some ProTools mixes by passing the mix out a couple of channels and
adding a bit more of a fill instrument or two, by doing this in the
analogue
world and recording the new mix back to PT. Do you think I'd be damaging
the material more in this way, than I would by trying to re-set a PT mix
which
was done completely in the analogue world and NOT in the box? (I'm on
Digi 192s clocked with an Apogee Rosetta, PT 7.2.1, HD2).

I'd be more concerned about laying in tracks to a stereo mix than I'd worry
about how many times you can do conversions. I had John Rice do some drum
tracks for me along with a stereo wave mix, which didn't require any special
reason to bring the mix back out to analog, unless you really are planning
to try to "mix" the new parts into the stereo mix. I've always hated doing
that, so I normally put up a couple of room mics and do playback through the
Mackie 1503s in the tracking room. Seemed to give a little better "analog"
feeling to it, not to mention being more of an inclusive part, rather than
an added part listening through headphones. But you have good enough chops
to have considered this.

Since I wondered off a little, I guess I should say that even with my MOTU
and the Tascam 38 and the DA38s a couple of passes each way wasn't a
particular problem. But it's not just asthetic appeal.

If the stereo mix wasn't done on your console, and you fly the tracks over
they will be different, so you won't actually be maintaining what you're
trying to add too.

Besides, David, what's wrong with the track. Sounds like Shania, btw.



--


Roger W. Norman
SirMusic Studio


"David Morgan (MAMS)" /Odm wrote in message
news:ScPak.265$HY.69@trnddc01...

"Scott Dorsey" wrote in message
...
Steve Byers wrote:
In your experience, assuming you use quality converters, how many
conversions (from analog to digital and back again) can typically be
performed before discernible quality loss is incurred?

I could do about thirty passes on the Prism AD-124 before it became
obvious. The Prism is by no means state of the art, though.

It's a fun test to make on your own hardware. Linearity issues get
very exaggerated.



There's some magic in the mix... it's just a little shy of some parts, and
re-creating the mix seems like a monumental task since it wasn't done
in the box to begin with.

Thanks,

DM

[philicorda[_6_]]

On Fri, 04 Jul 2008 08:18:13 -0400, Arny Krueger wrote:

"Ethan Winer" ethanw at ethanwiner dot com wrote in message

how many conversions (from analog to digital and back again) can
typically be performed before discernible quality loss is incurred?

I actually tested that with a $25 SoundBlaster card. I got a lot more
degradation than others report with "good" converters, but if you'd
like to hear the result here are links to the original Wave, after one
pass, ten passes, and twenty passes. Each file is about 2 MB:
http://www.ethanwiner.com/misc-conte...x_original.wav
www.ethanwiner.com/misc-content/sb20x_pass1.wav
www.ethanwiner.com/misc-content/sb20x_pass10.wav
www.ethanwiner.com/misc-content/sb20x_pass20.wav

Just for grins I made a composite stereo track of the left channel of
the original on the left, and the left channel of the 20th pass on the
right.

There are some expected things like a 4 dB loss at 20 Hz, and an 11 dB
loss at 20 kHz.

Wow, my ears must be getting old. If you still have the data around... is
it a sudden 11dB drop around 20KHz, or a gradual fall off starting at
-3dB at 15KHz or so?


This tells you what sort of fairly crude (by modern standards) converter
is in play. It's about 0.5 dB down at 20 KHz, while the good ones are
less than 0.1 dB down.

You say $25? My, times have changed! Back around 1996 the then current
SB64 cost nearly $200 and was very audible in one pass stereo, full
duplex.

Not expected was an approximate 0.3% frequency shift. IOW an artifact
of the original recording was shifted from 1003 Hz to 1000 Hz, and this
was consistent across the whole frequency spectrum.

Could this be a cumulative effect of sample rate conversions?
The files are at 44.1KHz, but most Soundblasters were fixed at 48KHz
clock only.


Of course ABXing the two was a slam dunk.

Not so with the pass1 file. I suspect the fun begins around pass 5. It
seems like a really good converter might do 10 passes and still pass an
ABX test.

[Arny Krueger]

"philicorda"
wrote in message
On Fri, 04 Jul 2008 08:18:13 -0400, Arny Krueger wrote:

"Ethan Winer" ethanw at ethanwiner dot com wrote in
message

how many conversions (from analog to digital and back
again) can typically be performed before discernible
quality loss is incurred?

I actually tested that with a $25 SoundBlaster card. I
got a lot more degradation than others report with
"good" converters, but if you'd like to hear the result
here are links to the original Wave, after one pass,
ten passes, and twenty passes. Each file is about 2 MB:

http://www.ethanwiner.com/misc-conte...x_original.wav
www.ethanwiner.com/misc-content/sb20x_pass1.wav
www.ethanwiner.com/misc-content/sb20x_pass10.wav
www.ethanwiner.com/misc-content/sb20x_pass20.wav

Just for grins I made a composite stereo track of the
left channel of the original on the left, and the left
channel of the 20th pass on the right.

There are some expected things like a 4 dB loss at 20
Hz, and an 11 dB loss at 20 kHz.

Wow, my ears must be getting old. If you still have the
data around... is it a sudden 11dB drop around 20KHz, or
a gradual fall off starting at -3dB at 15KHz or so?


Hmm, with different averaging I get:

0.2 dB at 10 KHz
0.0 dB at 12 KHz
1.0 dB at 15 KHz
2.5 dB at 18 KHz
5.0 dB at 20 KHz

and falling like a rock above 20 KHz which is probably where the 11 dB in
the former data came from.

[Ethan Winer]

Arny,

Not so with the pass1 file. I suspect the fun begins around pass 5.

I have all 20 passes, but I only put 1, 10 and 20 on my site. For giggles,
here's Pass 5:

www.ethanwiner.com/misc-content/sb20x_pass5.wav

If you'd like, I'll be glad to mail you a CD-R with the original plus all 20
passes. Email me from my site with your mailing address if you want it.

--Ethan

[Ethan Winer]

Phil,

Was it a SoundBlaster Live?

It's the SoundBlaster X-Fi that came with the Dell XPS tower I bought less
than a year ago.

It makes me think that converters are unjustly blamed for bad sound
quality far too often.

It kills me that the same "purists" who obsess about A/D/A conversion, and
turn up their noses at anything less than a Lavry or Apogee, are the same
folks who praise analog tape and toobs. So which do they want, warm and
fuzzy or transparent and pristine? :-)

Even more galling, most of these same people have no bass traps or other
room treatment. So they sweat over being 0.1 dB down at 20 KHz, and worry
about jitter well below the noise floor, but are blissfully unaware of half
a dozen 20 to 30 dB peak/null spans all below 200 Hz.

--Ethan

[William Sommerwerck]

It kills me that the same "purists" who obsess about A/D/A conversion,
and turn up their noses at anything less than a Lavry or Apogee, are the
same folks who praise analog tape and toobs. So which do they want,
warm and fuzzy or transparent and pristine? :-)

Aren't you assuming that all converters are about equally good?

I'm hardly anti-digital, but I've heard lousy-sounding digital (and
sample-data) gear.

[Peter Larsen[_3_]]

David Morgan (MAMS) wrote:

"Predrag Trpkov" wrote in message:

"David Morgan (MAMS)" /Odm wrote ...

Tell you what... If you think there's enough steel guitar in this
unmastered mix, I'll blow off the whole idea. I think it needs
more. http://www.m-a-m-s.com/MP3/BTZ/Heaven_In_Your_Arms.mp3

This mix sounds good to me as it is, but I can imagine it benefitting
somewhat from making the steel guitar as prominent as the acoustic
guitar fill-ins on the opposite side.


Agree.

I agree completely. The singer had a favorite steel 'lick' in the
last chorus and had me turn it up. That ended up sounding awkward to
me, but it's a reference point for where the steel should have been
throughout the entire song.

I kinda agree.

The whole problem here, was that I did
not have my little 6" Tannoys with me for the mix, so I mixed on the
big TAD's and headphones... and they apparently both lied to me
about the presence of the steel frequencies.

My Nokia has an spl meter. You need to bring one, the TAD's fooled you by
being at least 10 dB too loud and thus making guitars seem more obvious than
they are on this pair of Sennheiser HD430's on directly on my laptop.

You can do it both in and out of the box. Just align the tracks
visually as well as you can and it should be a breeze.

That's not so easy. In millisecond increments, 2 is not enough,
but three is too many. :-(

I've already busted my ass trying to do this inside the box. The old
"sum of the parts" principle (along with my unfamiliarity with the
inner workings of PT), has wasted too much time already. The mix,
while
pretty dense, requires that I severely turn down the steel guitar
track level in PT to make it fit with the mix. If I do this, I'm
losing all of the automation that I've written to that track. I'm
sure there's a way to turn down the whole track level and maintain
the automation, but I can't find
it. Back to the 'sum of the parts' thing... this is one of those
cases where a dB or so (or even less) of extra presence of the raw
instrument track, combined with the overall mix, could be too much.
But I had rather chance that, than take the 5 hours or so needed to
re-establish the analogue mix and maybe miss some of the magic.

It's not the easiest kind of arrangement to work on, so many busy
guitars. I like what you've done. How long did it take you to mix it?

There are three tracks of electric (one stacked rhythm and a solo),
four tracks of acoustic rhythm (with one stereo take being 2 of
those) , and
an acoustic solo; most all of if going through a multitude of outboard
processing and FX. That's an 8 hour mix, which includes the spot
editing and writing the track automation for about 22 tracks, IIRC.

I wish you had allowed a larger pan space for them.

What MCI console?

It's a 536... and if I do say so myself (as do the Blevins people),
probably the most pristine 536 left in the US. There's nothing on
it that doesn't work, but it's not wired for 'quad'. :-)

I'm pretty sure I'm going to take a run at this in the analogue world.
The guys at the studio are screaming "quality loss".... but I'm in the
same mindset as Mike Rivers' post of last night to this thread.

If you need to do it, then do it and be done with it. And yes, you should up
that steel guitar a bit over all. It is a charming mix overall, but too
monophonic for my taste. Do not change that now. Have it in mind for next
time. Also, while it is way beyond anything I have on the raw tracks site do
allow me to point in the direction of PL's mixes there anyway. Also because
of the stereophonic space issue. They are all static btw. You have a nice
and charming recording there Dave, one I will love to get over the counter
when released.

DM

Kind regards

Peter Larsen