|
Post by drbill on Jan 19, 2020 16:11:07 GMT -6
If anyone on the forum had a somewhat clean sounding summing box and could export one mix with that and one with digital summing this could be settled once and for all. With LCR panning to avoid differences in pan law and no random reverbs my guess is that it would prove digital and clean analog summing to be extremely similar, but I would love to be proven wrong. Most of us who were curious have been there, done that, years ago. No need to settle it for once and for all.
|
|
|
Post by lando on Jan 19, 2020 16:12:41 GMT -6
If anyone on the forum had a somewhat clean sounding summing box and could export one mix with that and one with digital summing this could be settled once and for all. With LCR panning to avoid differences in pan law and no random reverbs my guess is that it would prove digital and clean analog summing to be extremely similar, but I would love to be proven wrong. But the summing box needs a panning law. Or if it has none, which would be most usual I think, you need to disable it in your DAW, which WILL change the mix. This is what has skewed this kind of test on "other forums". Is that even if things are panned only hard right, left or center? Hard left should mean everything on one side and nothing on the other no matter the pan law? I truly don't know this stuff, so that is just how I thought it would be..
|
|
|
Post by lando on Jan 19, 2020 16:15:43 GMT -6
If anyone on the forum had a somewhat clean sounding summing box and could export one mix with that and one with digital summing this could be settled once and for all. With LCR panning to avoid differences in pan law and no random reverbs my guess is that it would prove digital and clean analog summing to be extremely similar, but I would love to be proven wrong. Most of us who were curious have been there, done that, years ago. No need to settle it for once and for all. Yeah, I kinda figured lots of people have done that, and I do believe your conclusions 100%, but it would be good and educating to hear the results of a test like that, specially for people like me who has no access to a summing mixer or a console.
|
|
|
Post by drbill on Jan 19, 2020 16:21:32 GMT -6
Most of us who were curious have been there, done that, years ago. No need to settle it for once and for all. Yeah, I kinda figured lots of people have done that, and I do believe your conclusions 100%, but it would be good and educating to hear the results of a test like that, specially for people like me who has no access to a summing mixer or a console. yup. There's no replacement for doing it your self and drawing your own conclusions. Unfortunately, to do a significant test across various genre's, styles, types of mixes and OTB summing devices takes a considerable amount of time and energy. Something not many have an abundance of. Then of course there's the inevitable. You screwed up the results. You screwed up the test itself. I don't like the material, it makes a HUGE difference on my songs - why didn't you do my songs. And just the general, all around, no good deed goes unpunished syndrome. So no....no thanks. But by all means, have at it yourself. We need someone willing to go downrange and stand next to the targets.
|
|
|
Post by stormymondays on Jan 19, 2020 17:13:23 GMT -6
But the summing box needs a panning law. Or if it has none, which would be most usual I think, you need to disable it in your DAW, which WILL change the mix. This is what has skewed this kind of test on "other forums". Is that even if things are panned only hard right, left or center? Hard left should mean everything on one side and nothing on the other no matter the pan law? I truly don't know this stuff, so that is just how I thought it would be.. That’s especially true if things are hard panned! Stuff on the sides will have a different volume than stuff in the center. Different DAWs have different pan laws and you can even change them (at least in Logic Pro which is what I use). You need to know what happens when you hard pan tracks in the DAW and in the su mming box. Here’s a quote from Steven Slate: “ I've shocked owners of hardware mixer boxes when they realize that the extra "width" they experienced was because their hardware pan law was 4.5 and the workstation was 3! ”
|
|
|
Post by javamad on Jan 19, 2020 17:46:43 GMT -6
popmann javamad So as to not derail UA thread. I am curious as to what the “math” being described as in error or truncation is. I assume (perhaps naively) that the sum in digital domain is an actual sum of all tracks on a per sample per bit basis, and this is a lossless sum (ie, as you sum 16 bit tracks you get a 16 bit output). So no dither, no truncation. This means the rounding error from excel isn’t as described, because rounding is truncation of the number of decimals, which would be to me the same as reducing bit depth. It seems that this means that while it isn’t infinitely accurate as analog, it’s exactly as accurate as the bit depth and sample rate you already recorded at...and if that’s not good enough, then it’s not summing that’s the problem but digital. I am quite happy to be educated about this if I’m not understanding correctly. I don’t know what math would be involved. Teach me, please. Not going to pretend to teach anyone :-) My assumptions on the process would be valid for mono signals too. So maybe the pan law and comments on this thread about depth are off the mark as to my original statement. So, is truncation the same as rounding? No. Truncation just lops off the smaller decimal places, rounding considers the value the have and performs a stipulated change to the decimal place before the cut off point, increasing it by one if the decimal place to be removed is 5 or over. A kind of rounding will take place when A to D conversion happens, which is what happens to an analog summed signal as we take it back into the digital realm. The conversion process will calculate to the best it can with the bit depth is has available the exact voltage if the signal n times per second, where n is the sample rate. When the signals stay in the digital realm, no new conversion happens, so no rounding of values, the values at each sample time are just added as numbers and its just linear math. Like I agreed on the LUNA thread, a lot of analog summing's effect on sound comes form inconsistencies of the electronics in the channels but I also believe that the rounding effect also plays a part. That said .. I am keen to see if the LUNA Neve summing does in fact sound as good as I am hoping. It does not even have to me mutually exclusive .. I could put it on the 8 groups I take out of the box and also sum them OTB with inserts, etc.
|
|
|
Post by matt@IAA on Jan 20, 2020 10:50:49 GMT -6
I think in the context of this discussion rounding vs truncation is a distinction without a difference -- or perhaps we could agree all rounding is truncation but all truncation is not rounding, but whatever. The net result is a decrease in precision with some rule applied for how to handle the tail end of the decrease (rounding by some rule, cutting completely, or last digit(s) random value allocation i.e., dithering).
Agreeing and expanding on your point to sort of set a waypoint in the discussion - when we do an analog to digital conversion neither rounding or truncation are really what's happening. We're not measuring the signal and then decreasing the precision of the measurement. We're constructing a new set of information based on samples, as you said. The process is destructive and irreversible - no digital representation can ever fully recreate an analog signal because you have inherently limited resolution and bandwidth. The level of destruction, error, information loss, etc is related to the amount of information we can capture at each sample (this is the precision of the measurement, the resolution) and the frequency of measurement (this defines the bandwidth or the upper limit of frequency we can capture without aliasing or error, the bandwidth). Also important to remember that sampling frequency sets the bandwidth - Nyquist rate tells you the highest frequency you can sample without aliasing (error or distortion). Long story short you need to sample twice as fast as the fastest frequency you want to capture.
So because of this, once we are in the digital domain we don't have the original analog signal any more. It's totally gone. So when we're talking about digital vs analog summing, we're talking about how we're going to sum these new digital signals. So any discussion about error, rounding, bandwidth, etc. is "baked in". If we agree that digital summing is sample addition, simple math, then it seems to me any question of rounding or truncation is not relevant to summing. If that's a problem, it's a problem that happens well before we are summing a mix - and going back to analog won't fix it, because the initial AD conversion is destructive.
When you say - "the rounding effect plays a part" - it seems like a contradiction. We already agreed that the rounding is done, so if you are doing AD then DA to analog sum and AD back again, "rounding" is a given. And in the sum itself there is no rounding or truncation, because you're summing samples (I think. I am not sure about this, it may be - likely is - more complex). The only way the "rounding effect" could play a part is if we're talking about digital vs analog on an end-to-end basis.
If the real question then is - ok, it's not about summing, but some summing systems have distortion characteristics we like and they sound better than distortion- and error-free digital summing - then it's not really a problem with digital summing. It's saying we like the distortion. Which is fine, and good - but that's really a different topic, yaknow?
|
|
|
Post by lando on Jan 21, 2020 15:19:37 GMT -6
Is that even if things are panned only hard right, left or center? Hard left should mean everything on one side and nothing on the other no matter the pan law? I truly don't know this stuff, so that is just how I thought it would be.. That’s especially true if things are hard panned! Stuff on the sides will have a different volume than stuff in the center. Different DAWs have different pan laws and you can even change them (at least in Logic Pro which is what I use). You need to know what happens when you hard pan tracks in the DAW and in the su mming box. Here’s a quote from Steven Slate: “ I've shocked owners of hardware mixer boxes when they realize that the extra "width" they experienced was because their hardware pan law was 4.5 and the workstation was 3! ” I didn’t know that, thanks!
|
|
|
Post by Guitar on Jan 21, 2020 15:41:17 GMT -6
Now I'm really wondering if my passive resistor summing mixer has a pan law different from what I have set in Cubase. Not sure how to check this (the analog one.)
|
|
|
Post by notneeson on Jan 21, 2020 15:45:01 GMT -6
Now I'm really wondering if my passive resistor summing mixer has a pan law different from what I have set in Cubase. Not sure how to check this (the analog one.) Even if you're summing mixer is full L/R dual mono with no pan pots, thus effectively handing panning duties back over to the DAW, I would expect some minor difference due to cross talk. But I'm no tech...
|
|
|
Post by Bob Olhsson on Jan 21, 2020 15:57:14 GMT -6
Summing is simply addition. The only complications are the need for gain changes. That's multiplication and then you are dealing with truncation, rounding and, when it gets converted to fixed point, dithering.
|
|
|
Post by Guitar on Jan 21, 2020 18:53:35 GMT -6
Now I'm really wondering if my passive resistor summing mixer has a pan law different from what I have set in Cubase. Not sure how to check this (the analog one.) Even if you're summing mixer is full L/R dual mono with no pan pots, thus effectively handing panning duties back over to the DAW, I would expect some minor difference due to cross talk. But I'm no tech... My summing mixer I designed with 4 mono inputs and a ton of stereo L R pairs. I guess you are right it would have to depend on the DAW still in this case. I can hear a difference, but my opinion is that it's not worth the headache of setting it up and printing it. It's a "small" difference to my ears. I also agree with PopMann that Mackies can sound great sometimes, surprisingly so. These days I try to get it all done in the box and am extremely happy in doing so. That may not be the point of this thread though. I am indeed interested in "hard answers" to the OP question as long as it is being discussed. I will try to understand.
|
|
|
Post by javamad on Jan 22, 2020 4:59:56 GMT -6
Interesting stuff on the pan law ... I will need to do som analysis on that myself.
|
|
|
Post by stormymondays on Jan 22, 2020 5:18:10 GMT -6
Well, an easy test would be to sum a mono mix!
|
|
|
Post by javamad on Jan 22, 2020 6:30:27 GMT -6
Found this online about pan law ... sounds reasonable
Many think (wrongly) that because they take an existing mix, change the pan law setting and get a wider/narrower image, that the pan law setting affects the perception of width in a mix. Really all they're doing is changing the volume of all their instruments depending on their panning position. Of course the width of the mix changes when you do that. But had you mixed with another pan law setting, for a given desired result, you'd have put the volume faders at another position yourself anyway, and gotten the same result in the end.
|
|
|
Post by Bob Olhsson on Jan 22, 2020 8:10:31 GMT -6
It's worth mentioning that even the tiniest gain change drop you into the world of rounding and truncation. Summing is not any kind of a special case.
|
|
|
Post by matt@IAA on Jan 22, 2020 9:30:32 GMT -6
Right - if there’s a problem it’s a problem with faders too.
|
|
|
Post by christopher on Jan 22, 2020 11:37:01 GMT -6
I’ve been watching some of those Eric Valentine videos, it’s an incredible insight to his workflow. He uses a DIY passive summing box with resistors. Somebody asked him in one of the QA’s why does he use it? His response was interesting, I’m not sure how factual of a response it is.. but it was something in reference to another video where they recorded guitar.
In that video he demonstrated how a 2 note chord sounds very different than recording each note of the chord through the same amp/settings and mixing them together on separate tracks- There’s a permanent separation when you record single notes. So in his logic, analog summing works the same way. I’m not sure if he ever tested this?, he does have the team members to be able to run any tests he wants, and they’ve done a lot of tests, double blind etc for many things, one example he says offline render is better than real-time in PT.
Here’s something I thought was happening during summing, but I’m not expert in electronics hahaha.. when you start patching channels together in analog, doesn’t that affect how each channel behaves? I mean the load of each channel is now connected to all the other channel’s circuits, right? Sort of like adding more speakers to a power amp? So I always thought that’s where analog summing has a sound. But is this thinking wrong?
|
|