A/B comparison between DAW SUMMING vs ANALOG SUMMING

danjpiscina said:

Don't conclude that just because it's only "0 and 1" it wont affect the sound. It does. Listen for yourself.

Click to expand...

A suite of 0 and 1 is simply a suite of 0 and 1...
I have APOGEE AD16 and DA16 converters, 2 MOTU 2408MK3 and 2 192i/o... I tested a lot of things and ADAT with wordclock is working like the others... They carry the 0 and the 1 to the point they have to without any issue when used with good fiber leads...
Anyway, it's out of topic...

EDIT: Just to be clear, I'm not talking about ADAT recorders... Bad converters even if some really cool sounding records where made with them...

danjpiscina said:

My point is that ADAT sounds horrible. It's a user-friendly and cheap way to expand your i/o, so for a pre-production/demo studio it's perfect. But it's just not professional and it sounds bad.

Click to expand...

Wow, is all I can say.  ???

Mark

lolo-m said:

I have APOGEE AD16 and DA16 converters, 2 MOTU 2408MK3 and 2 192i/o... I tested a lot of things and ADAT with wordclock is working like the others...

Click to expand...

That's interesting. So, you A/B'd the analog i/o with the ADAT i/o and didn't hear a difference? What are you using as your master clock?

danjpiscina said:

My point is that ADAT sounds horrible. It's a user-friendly and cheap way to expand your i/o, so for a pre-production/demo studio it's perfect. But it's just not professional and it sounds bad.

Don't conclude that just because it's only "0 and 1" it wont affect the sound. It does. Listen for yourself.

Click to expand...

Oh dear lord. Not in this forum as well! 

Biasrocks said:

danjpiscina said:

My point is that ADAT sounds horrible. It's a user-friendly and cheap way to expand your i/o, so for a pre-production/demo studio it's perfect. But it's just not professional and it sounds bad.

Click to expand...

Wow, is all I can say.  ???

Mark

Click to expand...

Yup better sell my two HD24s.. and disappear into the sunset then.......?!

Kingston said:

danjpiscina said:

My point is that ADAT sounds horrible. It's a user-friendly and cheap way to expand your i/o, so for a pre-production/demo studio it's perfect. But it's just not professional and it sounds bad.

Don't conclude that just because it's only "0 and 1" it wont affect the sound. It does. Listen for yourself.

Click to expand...

Oh dear lord. Not in this forum as well! 

Click to expand...

I totaly agree with you. End of discussion.

It's so funny how all of you guys take a technical criticism so personally. Almost as though I PERSONALLY insulted YOU and not a digital protocol.

All I'm trying to do here is make it clear that ADAT (lightpipe) is infirior to other digital protocols. I'm sorry that a lot of you use it but the reason I posted all of this is so that people would know that the ADAT outs had a great effect on the overall sound of this comparison.

P.S. to lolo-m: not sure why I forgot this, but ADAT lightpipe is a self-clocking protocol.

There's no need for flame war here...
I think danjpiscina talks about jitter issues in the ADAT protocol, but if the clock is good enough I don't think anyone would be able to notice any difference, since all it really does is carry 0's and 1's. And yes, ADAT can carry clock info, but you might as well use wordclock.

In any case, since the same converters are used in both exhibits, it doesn't matter if they're "average" or "excellent" or if ADAT is crappy or not. It's not like you will be able to hear any protocol/converter distortion, the source material is indie rock with tons compression and distorted vocals (yeah, the band wanted over-the-top flat-out sound, but I think I convinced them that it is a good thing to preserve some dynamics, too ;D )

Remember, the goal is to compare the summing!

I wasn't able to upload the tracks yet, had some *uckintosh problems. I'll try to do that tomorrow, though.

ytsestef said:

In any case, since the same converters are used in both exhibits, it doesn't matter if they're "average" or "excellent" or if ADAT is crappy or not.

Click to expand...

But in your ITB comparison, your outputs aren't converted from digital to analog, and then back to digital again, as they are in the passive summing mix. This is where the sound is heavily affected.

there's a very good thread on this subject over at prosoundweb:
http://recforums.prosoundweb.com/index.php/mv/msg/4918/0/48/0/

read Paul Frindle's comments

you are still trying to compare cashmere vs wool, but they both covered with mud!!! ( no offense )
i dont see any objective comparison ground here...

if you take your final ITB mix, and spread it out on a ssl/neve desk, it will sound crap... thats way on many tests, people
prefers ITB mix...
OTB mix doesnt mean matching levels to your DAW.... OTB mix means mix within your mixer's limitation with your outboard + plugins... and
which involves so many other elements that DAWs lack.... + high quality AD/DA is a must !

this is pretty much the same for basic summing mixer too.... at the end of the chain, you have a mic pre... which reacts differently then
your DAW... it may make things softer, edgier, duller, brighter.... you have to mix accordingly...

as some other pros pointed out...
you cant perform true A/B  test in this issue... but you can perform your best ITB mix, and perform separate OTB mix,
then compare both, and see which one you like...
they wont be same mix, but you will clearly hear which one you like , and why...

danjpiscina said:

ADAT lightpipe is a self-clocking protocol.

Click to expand...

danjpiscina said:

ytsestef said:

In any case, since the same converters are used in both exhibits, it doesn't matter if they're "average" or "excellent" or if ADAT is crappy or not.

Click to expand...

But in your ITB comparison, your outputs aren't converted from digital to analog, and then back to digital again, as they are in the passive summing mix. This is where the sound is heavily affected.

Click to expand...

The ITB thing is going OTB through the same converter (only using 2 channels, because it is already summed ITB) to get though the GSSL.

And to those saying the preamp colors the sound, I have to say that the results may be more similar-sounding than you'd expect. Then again maybe not.  :-\ Depends on what you expect, really. But you'll notice the difference is not in frequency response (surprisingly) but elsewhere.  :-X :-X :-X

I am trying very hard to find some time to actually upload the tracks ("aDSL" service providers in this country puts some really good use to the "a"   ). Should be back soon.

that bloody "a" strands for asymmetric, unfortunately
i have the same problem... up to 8mb on download, max .9mb on upload

ytsestef said:

I am trying very hard to find some time to actually upload the tracks ("aDSL" service providers in this country puts some really good use to the "a"   ). Should be back soon.

Click to expand...

feel free to use my idisk. the upload info is on the top post here:

http://www.groupdiy.com/index.php?topic=35303.msg432978#msg432978

you can upload as much as you like. that "public folder" is there for Prodigy Pro comparisons. All the best!

Dan P.

Thanks a lot, Dan!
Ok, the files are being uploaded right now.

http://idisk.mac.com/danjpiscina-Public?view=web

I used 24bit/44.1kHz in the end. The original multitrack was 44kHz/16 Bit, so I couldn't see any great benefit in using 48kHz, plus I get to save a little on bandwidth.

The song was recorded in a home studio by a hobbyist keyboard player in a low-budget setup. This means sm58 on vocals, vox tonelab amp emulations on guitars, DI Bass. Drums were recorded in a studio, but, because of the tight budget the drummer was only allowed two takes per song  : So, yeah, the source material is not that good. The mix was a rescue mission, and, for a demo, I think it turned out ok-ish (the "-ish" is for the stability of the drummer, snare flams in some sections of the song because the keyboardist edited some hits that would otherwise be way off-beat, although couldn't do anything about the overheads which contain snare too).

The band's site is the following:
http://www.theplayground.gr/main.html

Although for the moment, the tracks uploaded there are rough mixes not done by me.

ANALOG SUMMING:
The mix was done with the GSSL on. 16 channels in 2 ADAT pipelines went out of the DAW, into the A16 (44.1kHz). The analog outs of the A16 were fed into the passive summing box. From there into the 9K (+39dB gain), then GSSL and back in the DAW through a Mytek A/D.

DAW SUMMING:
Haven't touch a single knob in the GSSL or the DAW, just routed all the buses into a stereo channel bus. 2 channels in 1 ADAT pipeline went out of the DAW, into the A16 (44.1kHz). The analog outs of the A16 were fed into the GSSL and back in the DAW through a Mytek A/D.

I was shocked at the results of the test, because I was expecting some difference in frequency response and separation of the instruments.
The results showed some improvement in separation (especially that sawtooth-octave-synth thingy, which is way more present in the verses of the analog mixdown. This was expected. But the results showed less spectral differences between the two mixdowns than I thought. Ok, that was a surprise. But the biggest surprise is the fact that there are huge differences in dynamics and character of the transients, which is something I wouldn't think of!! To make a long story short, I find that the analog mixdown is more edgy, more lively, more rich and less washed-out than the daw mixdown. YMMV. I'd like to hear your opinion on this.

NOTE: Before even thinking of doing this test, I tried ITB summing and feeding the stereo mix through the the summing box, then the 9k. There were significantly less differences. So, this is the summing we're hearing, not the 9k.

for the umpteenth time, you are not hearing any differences between your passive summing and your DAW summing. Neither of these produce any significant distortion (passive resistor network vs. summing 32bit floating point numbers), that could be measured and taken into account as a variable in a simple audio test. They are well below anything you can hear, or for your A/D converters to capture.

The biggest, actually more likely the only significant variable in your shootout is the amount of D/A converter outputs used.

What you're saying really is that you like the cumulative effect of distortions - and there are many types you can actually hear - from the numerous D/A outputs you end up using with analog summing.

a DAW mix will take a hit from only one conversion, vs. 14 more stages in parallel with your passive summing.

Do with that information what you like and remember that psycho-acoustically more distortion is often better than less.

Kingston said:

for the umpteenth time, you are not hearing any differences between your passive summing and your DAW summing. Neither of these produce any significant distortion (passive resistor network vs. summing 32bit floating point numbers), that could be measured and taken into account as a variable in a simple audio test. They are well below anything you can hear, or for your A/D converters to capture.

The biggest, actually more likely the only significant variable in your shootout is the amount of D/A converter outputs used.

What you're saying really is that you like the cumulative effect of distortions - and there are many types you can actually hear - from the numerous D/A outputs you end up using with analog summing.

a DAW mix will take a hit from only one conversion, vs. 14 more stages in parallel with your passive summing.

Do with that information what you like and remember that psycho-acoustically more distortion is often better than less.

Click to expand...

We actually agree. I mean, conversion is a part of OTB mixing  : We are not hearing "the resistor network", we are hearing the "external summing method". I just wanted to point out that it isn't the 9k that much.

Kingston said:

for the umpteenth time, you are not hearing any differences between your passive summing and your DAW summing. Neither of these produce any significant distortion (passive resistor network vs. summing 32bit floating point numbers), that could be measured and taken into account as a variable in a simple audio test. They are well below anything you can hear, or for your A/D converters to capture.

The biggest, actually more likely the only significant variable in your shootout is the amount of D/A converter outputs used.

What you're saying really is that you like the cumulative effect of distortions - and there are many types you can actually hear - from the numerous D/A outputs you end up using with analog summing.

a DAW mix will take a hit from only one conversion, vs. 14 more stages in parallel with your passive summing.

Do with that information what you like and remember that psycho-acoustically more distortion is often better than less.

Click to expand...

quite simply, this is just not true. audio summed in the digital realm always sounds different than summing in the analog realm. for digital, this is to do with the internal routing of your sequencing software (if you listen, you will notice that, for example, pro tools sounds different from logic which sounds different to cubase!). the same is true of summing in an analog console, and therefore in a passive summing system. a Neve 8068 will sound different when compared to an SSL4000 just because of the way they sum.

ps. Stefanos, i listened to both mixes, A/B'ing them. It's pretty interesting! Overall, I prefer the low end on the OTB one, but the ITB one has more bite. i prefer the OTB in the verses and ITB in the choruses! somehow there's a more musical-sounding punch in the snare during the verses. something to do with the way the GSSL reacts? maybe it's hitting it harder OTB?

overall there is a more cohesive mix OTB. the separation in the instruments and clearer and better defined than it is ITB. good times!

danjpiscina said:

quite simply, this is just not true. audio summed in the digital realm always sounds different than summing in the analog realm. for digital, this is to do with the internal routing of your sequencing software (if you listen, you will notice that, for example, pro tools sounds different from logic which sounds different to cubase!). the same is true of summing in an analog console, and therefore in a passive summing system. a Neve 8068 will sound different when compared to an SSL4000 just because of the way they sum.

Click to expand...

sigh.

please amuse me and present accurate measurement data to prove this.

Otherwise I will time and time again call this bullshit. Telling a four year old Santa does not exist, I do realise the futility (and cruelty) of my effort.

The digital part of your argument is an increasingly popular myth (impossible to kill, ref. 'belief systems'), and any properly set up nulling test will show you why. I know this is somehow completely impossible to do and hence rarely seen, but we all know how little actual engineering skills your average studio hack these days has.

The rest of your argument (pulled out the neve/ssl/analog rabbit out of the hat all of the sudden) introduces a plethora of variables in the form of mix bus buffers (imperfect analog amps), preamps (more imperfect analog amps), EQ stages (guess what?) and AD converters, all sprinkling little IMD, THD and noise over your audio. Of course they are going to sound different compared to perfect DAW summing!

Your predictable next step is trying to confuse me with the magical realm of floating point math, but ahha! that happens to be my day job - working with magic elves that is. I will leave you with this little bit of information: summing several thousand 32bit floating point numbers result in rounding errors that rear their ugly head somewhere down -180dBfs.

danjpiscina said:

if you listen, you will notice that, for example, pro tools sounds different from logic which sounds different to cubase!

Click to expand...

GEARSLUTZ.........that's where this belongs...and that's also where a post exist proving the DAW's you mention above null each other out.