R/E/P Community

Please login or register.

Login with username, password and session length
Advanced search  

Pages: 1 ... 7 8 [9] 10 11   Go Down

Author Topic: External summing of DAW mixes  (Read 77241 times)

cerberus

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2651
Re: External summing of DAW mixes
« Reply #120 on: November 16, 2008, 08:34:18 pm »

Bill Mueller wrote on Thu, 09 October 2008 12:26

I will state the simple fact yet one more time. If you invert and sum your ITB mix and your OTB mix and they perfectly cancel, son yer trippin' if you think one sounds different than the other.

bill; to "perfectly cancel" means: a bit-identical copy. otherwise the data is  different.
if someone can hear that difference in a double blind a-b test, then that is
the valid data worth investigating. not a white paper which tries to
prove that a person who "hears things" is "tripping". imo, we
ought to try and explore such phenomena rather than to
question the veracity of the observations and claims.


Vertigo wrote on Mon, 06 October 2008 12:43

I really find the Nuendo mix bus to be terrible in comparison...

lance;  i assume that you are montoring through a dither? since it would be
incorrect to judge sonics through a truncated output. if you find a sonics
bug in nuendo (or in cubase), then i would be interested in presenting it
to steinberg's beta testers and to steinberg's engineers in hamburg. t.i.a.

jeff dinces

Peter Weihe

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 371
Re: External summing of DAW mixes
« Reply #121 on: November 19, 2008, 03:10:14 pm »

Hi all,

here is another result of the 0-test:

It can be interpreted as a support for William's and Terry's theory to better use only one kind of Mic Pre Amp on one production for more consistency in the sound of a record.
And the test clearly shows why people always hear differences in the sound of a Mic Pre amp inside a console and a single unit outside a console.

1.) The leftover of two identical line-stages in a row is almost always exactly 6 dB higher than the leftover of a single stage. That means the artifacts are doubled!

2.) When a stage with a strong character and a high level of leftover (like a 1084
-18dB with my specific program) is followed by a relatively neutral line stage ( leftover -42,5 dB) the level of the 1084's leftover is only increased by 0,5 dB and the sound of the leftover stays pretty much the same.

That may sound only logical and maybe naive to mention but I was surprised about this clear result.

I was testing what line stage would be the best buffer to preserve the sound of my V72s or in other words cause the least alterations. The best of course was no second stage but if you want to ride a fader or need a summing amp there must be another stage.

I found that there were two solutions:
1. find the most neutral line amp.
2. take the same ( in this case a second V72s)

I remembered the first summing stage that Manfred had built for the summing test.
I had chosen it as the most neutral in a double blind test some years ago.
I borrowed the little mixer from a Tonmeister who had it in his studio and again for me it was the clear winner in " What line stage changes my V72s sound the least."

The other even more charming solution was the second V72s but I noticed that certain typical aspects in the V72s sound increased.

I repeated that test with 38 combinations of mic pre amps and line stages and the results were the same to my ears.
1084 remained 1084 when it was followed by a second one.
The same was true for API , Helios, GML ....

However in some cases the typical sound of certain units was simply sounding exaggerated to my ears with some signals when I added a third unit of the same kind to the signal chain.

So I wondered whether this could be heard or seen in our 0-test.

Test:

A.) First I made a run with Manfred's first summing amp and I was very happy to see and to hear that the leftover proved that this is the most neutral stage in the test so far. ( Only the Haufe and Malotki transformers had less leftover).

GML 8200 had been the one with the lowest leftover before ,
it showed -33.9 dB with the program I used.

Manfred's stage showed -42,5 dB

Again the 0-test corresponded with the results of my subjective listening tests.
I have to admit that I had to give up one of my dogmas because this summing amp of Manni is based on ICs .( Ahemm)  It's a symmetrical design with servos, no condensers.

B.) Next I used two Brent Averill 1084 in a row and compared the leftover to the one of only one unit.
One 1084 showed -18dB
Two units showed -11. 3 dB
All aspects in the character of the leftover increased. It was the same sound, only louder and more intense.

Next were 2 API 560.
I have always heard a clear difference between the sound of the 560 and the 550b on bypass.
But I was really surprised how big the difference of the leftover between those two units really was.

The leftover of the 550b sounds sharp in the upper midrange. Level -19 dB.

The leftover of the 560 has no significant treble but a strong low midrange boost, Level - 25 dB
They could hardly sound more different.

Two of the API 560 in a row doubled the level of the leftover and kept the sound character. -19 dB . Again the level was doubled.

William Wittman has posted several times that he prefers to use one type of API Eq
only in order to get a consistent sound. This little test shows why.

I repeated the test with Helios, GML, Manfreds mixer and always the level of the leftover increased by 6dB.
(Edit) :
>Helios through Helios Bus and post fader line driver
>GML 8200 through another channel of GML 8200
>Manfreds Mixer left bus - through - right bus.
>API 560 -API 560
> The level of the leftover with each of those chains was +6dB in comparison to the leftover of a pass through a single unit.
The level increased but the sound peaks had the same specific character as a pass through a single stage.
(End of Edit.)


C.) However Manfred's summing stage increased the level of the 1084's leftover by 0,5 dB only. The 1084 was the dominating character. Or in other words a neutral stage like that can transport the sound of a unit with a strong character without significant alterations.

D.) Then I combined a 1084 with an API 550b. Both units individually had a very strong significant sounding leftover.
The sum combined both of their characters and with the strong low end boost of the 1084 and the strong high midrange boost of the 550b the leftover of the sum was all over the place only a little less than 6dB more than one unit.


One thought struck me.

If an unexperienced engineer recorded a whole record through one console filled with one sort of line stages and the same Eq on each channel, chances are high that the character of the sound of all individual signals and even the buses are pretty much the same concerning the alterations the board's circuits do to the sound.

If now the mastering engineer has to figure out where to make his boosts and cuts,
chances are much higher with this recording that he could find certain frequencies
that were consistent on all instruments than it would be on a recording that was made through all different kind of circuits.

I am not saying that an experienced engineer cannot make use of this variety of different sounding circuits in order to purposely shape his individual tracks.

I am sorry, if these findings may sound too obvious to the more experienced engineers among you.

Thanks ,
Peter



















I
Logged
Peter Weihe

compasspnt

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 16266
Re: External summing of DAW mixes
« Reply #122 on: November 20, 2008, 03:13:32 am »

Great work, Peter.

I am slowly digesting this now.
Logged

Peter Weihe

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 371
Re: External summing of DAW mixes
« Reply #123 on: November 20, 2008, 02:25:07 pm »

Thanks Terry!

I made a striking new experience with the 0-test when I played it to students at the University Of Hannover last weekend.

One of them asked me if one could really hear differences between the sounds of different audio designs. I said sure you can and then he asked whether this could not simply be a result of imagination and that he had heard that it would be impossible to prove.

I loaded the 0-test from my hard-drive and explained how it worked.
They first heard the mix, then how the pass through the PT converters changed the sound, next how two runs through the PT converters nulled when one phase was reversed.
After they had listened to the leftovers of various examples of units that they knew from the university studio one of them said and the others agreed:

"Should we then not better start with one kind of pre amp and EQ and first learn what it does and experience how our recordings will sound in the end?
We don't know the different brands well enough to decide which one to use for what instrument." Bingo!

They suddenly really showed an enormous respect for the effect one pass through
one or the other kind of audio design had on the signal.
Before that experience on all other workshops before most of the students were really burning to try all of those nice outboard units that they probably had read about on gear slutz.






Logged
Peter Weihe

compasspnt

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 16266
Re: External summing of DAW mixes
« Reply #124 on: November 20, 2008, 05:58:48 pm »

Seduction is a Powerful Force.
Logged

Adam The Truck Driver

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 458
Re: External summing of DAW mixes
« Reply #125 on: November 25, 2008, 07:47:03 pm »

I know that I want to use analog, either an ATB now with meter bridge, or Elan II TT, but might find the 8816/8804 more practical & I'm expecting to have a very nice selection of outboard preamps, CLs & EQs...I do know that I only want to use the DAW as a
multitrack recording machine and as a digital efx box with all
dynamics and EQ processing done otb. That is what I want, but it doesn't really matter does it? It's an impulse...sorry.
Logged
Adam Brown

No File Selected.








MDM,

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2305
Re: External summing of DAW mixes
« Reply #126 on: December 04, 2008, 08:52:34 pm »

Peter,
the leftover signal from the tests does not necessarily mean harmonic distortion etc.  It could be that the phase shift  is  due to internal coupling caps or transformers.

This means that the high leftover signal is not necessarily an indication of poor performance with audio, which is full of phase rotation etc. In nature so its not super critical.

But if the phase distortion is erratic and complex like in high feedback circuits then it is a pain to the ear for sure
Logged
I must study politics and war that my sons may have liberty to study mathematics and philosophy .. in order to give their children a right to study painting, poetry and music.
John Adams (1735-1826) 2nd President, United States

Deuce 225

  • Jr. Member
  • **
  • Offline Offline
  • Posts: 78
Re: External summing of DAW mixes
« Reply #127 on: December 07, 2008, 09:01:04 pm »

Like Terry said a few posts back, I am still "digesting" this thread. At the risk of sounding like a complete dope, I will admit I have read this thread several times and I'm still not sure I understand the key conclusion(s) and how best to apply the principles to improve a mix. On one hand, it seems to make an argument for tracking OTB on a console with "like" pre's and EQ's.  It also seems to suggest the possibility of improving OTB tracking by parallel chaining like pre's i.e. two GML's or 1084's etc...

Lastly it seems to make the case for performing as few AD/DA conversions as possible. The original post was entitled "External Summing of DAW mixes".  To apply "Peter's principles" to summing is where I get a little lost.  Obviously to SUM OTB will require additional AD/DA conversions.  So...unless there is "hearable" compensating value for performing the additional AD/DA steps it would be hard to build the case for summing OTB.  I have been considering purchasing an Inward Connections Mix690, but it sounds as though a better strategy is to find a good used console.  I welcome other perspectives.

Thanks Peter for all your work.
Logged
Best,

Tim Cochran

Peter Weihe

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 371
Re: External summing of DAW mixes
« Reply #128 on: December 10, 2008, 05:14:00 pm »

Deuce 225 wrote on Mon, 08 December 2008 03:01

... I will admit I have read this thread several times and I'm still not sure I understand the key conclusion(s) and how best to apply the principles to improve a mix.


Hi Tim,

I must admit that I have posted lots of results but purposely left it to the reader to
come to his individual conclusion. I will post some opinions and conclusions that I
have heard from engineers who discussed the findings of this thread with me.

I am afraid there are no universal conclusions beside the ones that the O-test proved that -

1.) there is and must be a significant difference between a mix ITB or an analogue mix with all levels and panning set to the same values. -

2.) It showed that the extra DA conversion alone changes the signal significantly ( all listeners felt that the change was to the worse with the PT converters) plus

3.) each extra circuit that the signal has to pass through changes the sound more or less, each with it's specific footprint.

But that alone doesn't say which way is objectively better for what kind of style and mix.

Bill has proposed the 0-Test because he hoped that it would bring us an objective, empirical answer instead of the common exchange of subjective opinions, which are necessarily based on different recordings, different converters, outboard equipment, rooms, monitors...
I purposely tried not to value my findings too much but just report my observations and leave my taste aside in order to stay as neutral and dare I say as "scientific" as  my little private setup allowed me.

If the test made any sense than it would have to mirror the individual findings of those of us who have tried various setups and grades of ITB or OTB scenarios and evaluated them by listening because in the end our individual ear-brain-socio-cultural history and style dependant taste is all that counts.

We have to feel comfortable with our gear, the work flow and we have to like what we hear.

I am happy that I kept my notes from the first double blind listening test that I took part in with some of the units that I included in the 0-test, because those notes mirror exactly what I heard in the leftover of Bill's 0-test.

Quote:


On one hand, it seems to make an argument for tracking OTB on a console with "like" pre's and EQ's.  It also seems to suggest the possibility of improving OTB tracking by parallel chaining like pre's i.e. two GML's or 1084's etc...


I have got lots of responses for the tests and people still come to completely different conclusions.
First this is again what I have found:

My 0-test with inserting two or more units of the same model in the path showed that:

4.) the level of the leftover raises by almost always exactly +6dB when to stages of the kind are put in series. The sound character of the leftover stays the same as with one unit but it gets more intense.

5.) when I put two different sounding units in series each with a distinctively recognizable sound in the leftover the sum showed a mix of the two characters.
For example the leftover showed the low end boom of the BA 1084 plus the high mid bite of the API 550b.

6.) I do interpret that as a prove for the observation that a module sounds different
inside a console with all cables, transformers and similar circuits following in the signal path than a single module outside the console.

Some of the engineers who listened to the files liked it a lot, when they heard the sound of two chained APIs oder 1084s and others thought that some signals were  getting too much of that character. Others felt that it was great when there were three API op-amp transformer circuits in series and said that 4 of them were too much for their taste...

7.) My conclusion is that this test supports William's and Terry's theory that you get a more consistent sound when you use just one kind of mic pre or eq or record everything through one console filled with the same type of Eqs.

Now some friends of mine come to the exact opposite conclusion.
They take this as a prove that they can purposely carve a more transparent mix when they record with different mic pre amps, use different eqs and compressors
and choose each unit for a specific character that they like to add to each instrument.

Quote:



Lastly it seems to make the case for performing as few AD/DA conversions as possible.
The original post was entitled "External Summing of DAW mixes".  To apply "Peter's principles" to summing is where I get a little lost.  Obviously to SUM OTB will require additional AD/DA conversions.  So...unless there is "hearable" compensating value for performing the additional AD/DA steps it would be hard to build the case for summing OTB.  



Yes, everyone among the listeners agreed that each extra conversion degraded the sound and stole detail.
But again there are different conclusions:
Everybody seems to  set the individual threshold for the amount of degradation on one side and the benefit of their outboard gear on the other side differently.

Some examples:

A.) Ross wrote that he does all overdubs through a Hedd 192 converter and monitors through the Cransong Avocet. He doesn't like the sound of an extra pass through the PT converters and prefers to stay digital. His master insert goes via AES to the Cranesong Hedd.

B.) I came to the same conclusion since I bought the Layry Gold. I record all overdubs through the Lavry and I monitor through a Lavry Blue and my console or through the Avocet.
For mixes in my room I try to rent the best converters I can get and let the engineer
mix through my analogue console.

C.) A friend of mine loves to mix on his SSL Duality console and his huge collection of outboard gear to a 1" stereo analogue machine . He uses the PT 192 converters although he knows from his own tests that there are better ones. But he feels that his gear and his work-flow outweighs the degradation of the extra pass through the converters.

D.) Other colleagues of mine are looking for better converters and depending on their budgets you can find a variety ot the usual suspects.

E.) One of the engineers, a tonmeister, who took part in our tests over a long period of time uses different setups for different styles of music.

He loved the API modules for rock 'n roll and hated them for classical music.
When he recorded rock bands at our university he borrowed my API lunchbox and some vintage discrete Siemens units.

For his classical recordings he uses a very puristic setup and stays in the box after the first AD conversion with high class converters.

For score mixes he likes to get some character and  he loved the sound of the 1084 in the master insert. But he needed recall and preferred the EMI Curve Bender with it's step controls.

Yesterday he told me that he would love to mix his Jazz recordings on my console.
He was very open minded during our tests, listened carefully and felt that each style of music benefited from a specific setup.


One more conclusion:
I believe that if one decides to work ITB one should try to get the best DA- converter one can afford, because those are the lenses we look through and base our decisions on. I see lots of posts from people with shitty AD and DA converters ( on other forums of coarse) who prefer to buy the seventh brand of mic pre amp and Eq " for more color" instead. Somebody should go over and tell them.


Quote:

 I welcome other perspectives.

Thanks Peter for all your work.



Thank you for your feedback,

best Peter
Logged
Peter Weihe

cerberus

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2651
Re: External summing of DAW mixes
« Reply #129 on: December 12, 2008, 09:21:45 am »

peter;

would you please remind people to dither when reducing bit depth?
that would apply to any d/a conversion that has been processed
in any way by the daw, including a simple gain change. it
doesn't matter if your daw is floating or fixed, since the
internal bit depth will always be greater than 24 bits.

t.i.a.

jeff dinces

Trumpetman2

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 357
Re: External summing of DAW mixes
« Reply #130 on: December 17, 2008, 02:58:43 pm »

wwittman wrote on Mon, 06 October 2008 18:55

Bill_Urick wrote on Mon, 06 October 2008 07:35

I have tended to dismiss this, but am reconsidering.

Are you doing it? Do you feel that it makes a significant difference? What are you using?

Thank you.




Huge difference.

I've been using the Dangerous 2 Buss


The BEST thing is to come out into an actual console, naturally.
But given the choice between mixing entirely ITB and using the summing box, the summing box wins by a lot, for me.





What about using a summing box after the mixer?  Would it "improve" the quality of the mix, or would the mixer impart its signature to everything after it...?
Logged

zmix

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2828
Re: External summing of DAW mixes
« Reply #131 on: December 18, 2008, 09:59:04 am »

cerberus wrote on Fri, 12 December 2008 09:21

peter;

would you please remind people to dither when reducing bit depth?
that would apply to any d/a conversion that has been processed
in any way by the daw, including a simple gain change. it
doesn't matter if your daw is floating or fixed, since the
internal bit depth will always be greater than 24 bits.

t.i.a.

jeff dinces


Jeff,

I'm confused by your comments...  Has Peter been instructing people to send you files which have been truncated without adding dither?

You might be relieved to learn that dither is added after truncation, and that most 24 bit D/A converters have enough residual noise ( above -144dBfs ) to dither the signal.

tunetown

  • Full Member
  • ***
  • Offline Offline
  • Posts: 108
Re: External summing of DAW mixes
« Reply #132 on: December 18, 2008, 05:05:02 pm »

Hi,

Can I ask someone to clarify the difference between a DA conversion of a summed ITB mix and sending 24 channels of DA conversions to a desk and summing OTB. I assumed this was the same process.

Is it the reconversion back from the desk to ITB that is being mentioned as the extra conversion process and therefore loss of quality. I haven't noticed this to be the case.

Great topic. Thanks for all your input.

Cheers
Logged
Peter Houghton
Tune Town

cerberus

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 2651
Re: External summing of DAW mixes
« Reply #133 on: December 27, 2008, 08:06:54 pm »

zmix wrote on Thu, 18 December 2008 09:59



I'm confused by your comments...  Has Peter been instructing people to send you files which have been truncated without adding dither?
hi chuck,  no. of course not.
Quote:

You might be relieved to learn that dither is added after truncation,

by definition: that woudn't be a dither; but mere noise.
Quote:

and that most 24 bit D/A converters have enough residual noise ( above -144dBfs ) to dither the signal.


and i don't get how that is relevant to production? once signals are run through more
dsp, they aren't 24 bit signals anymore. so  any low-level digititus which may have
infected the signal  due to lack of proper dithering could become multiplied.
perhaps you would think it is ok to mask it with noise, but sound is
fractal; a room has a noise floor, and there can still
be music inside of that. likewise with all gear.

maybe you like the noise floor of some analog gear even
though it is much higher than -144db? you might
find it euphonic. but  dither is not necessarily
euphonic by itself. a dither's color is
incidental to its effectiveness

it's not that dither noise sounds
good, but that it will work
(when the process is
applied correctly).

if the self-noise you refer to is analog noise, then it would be too late then to talk
about  dithering; that had to happen in the digital domain. as for digitally
induced "self-noise" ?  i think  that could be from truncation, or other
bad  maths? so that wouldn't likely turn out euphonic either.

jeff dinces

Bob Olhsson

  • Hero Member
  • *****
  • Offline Offline
  • Posts: 3968
Re: External summing of DAW mixes
« Reply #134 on: December 28, 2008, 03:07:22 pm »

Rant on:

Adding noise is not dithering and noise added after truncation is most certainly NOT dither. Dither sounds like noise however noise does not not dither unless it is added prior to bit depth reduction and meets very specific spectral, level and randomness criteria.

Adding to the confusion is the fact that dithering requires significant computational overhead that cuts down on the number of features a dsp device can offer. Instead of being honest about this, audio marketers have made excuses like "dithering is optional for use only when you hear a problem,"  "it's self dithering" or "it's such a low level you can't hear it." Sadly even many programmers want to believe this B.S. bad enough that they don't question it. Just as with the misuse of integrated circuits, it's common to just throw building blocks of code together that have fundamental flaws when examined closely.

This has been compounded by flat out broken dither routines from time to time that sounded even worse than truncation.

Rant off.

Now back to your regularly scheduled programming. Laughing Rolling Eyes  Shocked
Pages: 1 ... 7 8 [9] 10 11   Go Up