The plights of AD / DA "re-conversion"

[Donny]

At the studio today i decided to do an experiment to check how much a source is degraded upon continuous ad/da conversion. I took a sample of work i had and went out via the Digi 192's DA and back in through its AD's. On my first trial i sample aligned up the original and then converted to see if i could identify differences in the wave and i was just shocked to see how on some transients there were immense differences. I flipped the phase to see how much they would cancel and obviously the difference was quite noticable . I then decided to keep on reconverting and on each step i could not believe how much the wave was changing when you zoomed down to sample level. What a shock this was!!! I expected change but not this obvious. On latter conversion stages audible differences became much more noticable that even a client started to recognise.

I then decided to change to my lavry blues to see how much difference this made. The difference was there but not as huge as the 192's but nevertheless it was audibly obvious after a few re-conversions.I started to think how much this degradation must affect mixing during a typical project that uses outboard. What a bummer! Just wanted to see any opinions or other experiences on this matter....

[RABE]

ABSOLUTELY TERRIFYING. Ihttp://gearslutz.com/board/images/smilies/errrr.gif guess we alll have to mix ITB. Sigh...

[thethrillfactor]

Originally Posted by Donny

Just wanted to see any opinions or other experiences on this matter....

Easy solution...


Mix on a console and stay analog till the end.
(actually this would make a great slogan on a Tshirt for AES).

[swan]

This is extremely interesting. This might be noobish, but I always wanted to know what would yield the best results: bouncing inside cubase/nuendo OR sending going analog out of Rosetta, and back into analog ins, OR going spdif out of rosetta to spdif in of rosetta when bouncing my mix....

any insights?

[Dirty Halo]

You have to factor in what you may gain from one extra conversion. I mean, how much was just one conversion and if you factor in what you get by running something through a nice piece of analogue gear, then back in, is that added "plus" to the previous dry signal out-weighed by the "minus" of one conversion.

I usually don't bounce and bounce, it's usually a tracked signal, then maybe one out to a good piece and back in, from there it is mixed out through a console for good.

So, to the gentlemen who did the orignal experiment, was that first bouce actually audibly different?

Exactly how many bounces until audible artifacts could be heard?

This is not a challenge, I really would like to know and I appreciate the experiment by you.

Sooo...what are people's opiniuon's on the above: Does one coversion outweigh what you gain by running something through a yummy piece of real analogue gear, be it a compressor, reverb, whatever?

-a

DIRTY HALO www.dirtyhalo.com

[XSergeantD]

Originally Posted by Dirty Halo

Sooo...what are people's opiniuon's on the above: Does one coversion outweigh what you gain by running something through a yummy piece of real analogue gear, be it a compressor, reverb, whatever?

Yes.

For this purpose, we're all tracking into a DAW.
Therefore you're either mixing in the box or you're going out to your console/OB gear changing the waveform and printing to analog or back to your DAW. Thats 2 or 3 total conversions in comparrison to one conversion for mixing ITB. Worth it, yes in my opinion

[Dirty Halo]

Originally Posted by XSergeantD

Yes.

For this purpose, we're all tracking into a DAW.
Therefore you're either mixing in the box or you're going out to your console/OB gear changing the waveform and printing to analog or back to your DAW. Thats 2 or 3 total conversions in comparrison to one conversion for mixing ITB. Worth it, yes in my opinion

I'm sorry, do you mean:

1. "Worth it, yes" that you would rather NOT go through hardware and back in or...

2. "Worth it, yes" you don't mind the extra conversion given what you get from the outboard gear?

I'm a bit dense tonight...need help

-a

[Dirty Halo]

How many of us are printing FX?

Most compression and things like that, I will just insert on mix, but I often print some compression or re-print for "a sound", then there's the "out there" FX that I print, cut, tweak more, etc.

Anyone dedicate verbs to print? Etc.?

-a

[sage691]

I agree with Thrill Factor.

First of all, record in a great sounding room (wood ceilings and wood floors) to capture that natural reverb and tone. ISO booths are good for rap, but SUCK for singing and just about everything else IMO! Great acoustic guitars and pianos do not have their bodys made out of sheetrock or carpet!! What do you think a guitar would sound like with a sheetrock body ? Think of your recording room in the same way.

If you can get this first step right, then record into your DAW using great A/D converters. Even an old stock Apogee AD800 will do if your room and your analog mic pre is up to par. NOTICE I said ROOM first, and MIC PRE second! A greta mic pre capturing a shitty sheet rock sounding room is going to yield a nicely polished turd of a sound, and it won't be the mic pre's fault.

Then, if you can mix the outs of your DAW into a great analog console and record that to 1/2" tape if possible, or at the very least Lavry A/D to a Masterlink.

I know several pros (one BIG TIME) who do it exactly this way when they cannot track to 2 inch tape. Again and again they have told me that the ROOM matters more than ANY of the gear! One said he would choose 16 bit adats and mackie mic pres if he had to choose between a sterile room w/ top gear or a great room with 2nd rate gear.

I think many of us worry way too much about the digital tech stuff, and forget about the most basic things of all... i.e.-- great room and smooth, soulful performance!

[squeegybug]

Originally Posted by Dirty Halo

Does one coversion outweigh what you gain by running something through a yummy piece of real analogue gear, be it a compressor, reverb, whatever?

I gladly send through conversion to get to my analog bus equipment. No doubt it is worth it to me.

I *don't* send signals 5 or 10 times through the converters....

Steve

[XSergeantD]

Originally Posted by Dirty Halo

I'm sorry, do you mean:

1. "Worth it, yes" that you would rather NOT go through hardware and back in or...

2. "Worth it, yes" you don't mind the extra conversion given what you get from the outboard gear?

I'm a bit dense tonight...need help

-a

Sorry, I don't read what I type before I hit 'submit' so sometimes things come over wrong or ambiguous, so my answer is that you got it right

























OK, OK, answer # 2 is right. And as everyone here will probably note and mention that step one(or your current step) is most important than your next step cause it's more difficult to go back and fix a previous step. Remember "we'll fix it in the mix" went to "we'll fix it in mastering" and unfortunately now it's "we'll fix it in marketing"

Importance chart I'll make up now:
talent>instruments>engineer>room>microphones>mic pre>pre-printing gear (comp.,...)>converter/machine>gear>console>mix medium (1/2", A/D conv)> mastering engineer>Mastering eqpt>A&R>...

Ok, so I left some stuff out, but someone else can fix and/or add some stuff or put stuff in a better order

[Kiwiburger]

My 2 cents:

The degredation of d/a & a/d is nothing compared to bouncing to and from mag tape. And yet some of the best sounding records of all time had numerous tape bounces. I actually think this can give depth, that is missing from non-bounced digital stuff.

A good analogy is comparing tape or analog delays with digital delays. Most people much prefer the degraded sound of tape or analog delays. I believe this is because acoustic sound degrades in air - the further away, the more degraded. And therefore delays or distant sounds are more natural when allowed to degrade.

I also think that digital reverb (which is a whole bunch of delays) also benefits from having the sharp edges dulled by some smeary analog circuitry.

So for 'layered' send effects, especially with a time component that gives the pschoacoustic illusion of distance, it's probably very beneficial that the d/a & a/d process knocks off some transients.

Another important thing to think about is the fact that not all digital audio waveforms as we see them on the screen are valid representations of a continuous acoustic waveforms. Especially if any processing has occurred. The D/A converter will convert the data into an actual analog waveform - which is a very different thing. In the process, it can be expected to smooth out the waveform and, it's little wonder that when this is re-digitised by the A/D, it will appear slightly different on our screens. But different may actually sound better.

I'm also of the opinion that most converters sound worse when the signal is too hot or approaching 0dBFS. (I know other people don't believe this - it does depend on the converters you have). But it seems to me that few people are using Lavry Gold or similar high end converters for this out-of-box experience. The fact that better converters exist means that my Lucid and Benchmark converters aren't quite as good, and therefore there should be audible and visible differences in the waveforms.

I have no doubts that ITB digital streams offer the least change in a signal. But change can be good.

[RKrizman]

Originally Posted by JYoder

There are definitely issues with conversion, but this sounds like there's either a calibration problem or the soft limit is on....

Sounds like some other problem to me too.

-R

[u b k]

that's a lot of commentary on how the waveforms looked but not much about how they sounded. i can't imagine what all the transient stuff is about, other than to ask how your levels are calibrated and whether the input is recording the same level that the output is giving.

also, you are assuming de facto that different = worse, that change = degradation. bounce a track from one place to another on a studer and the waveform will most definitely look different, but is it worse? i like a bounce or two on tape for some tracks, it adds a nice haze, sweetens things up.

just be real about what you need. in most cases, using outboard requires one d/a conversion. if you're summing itb, add one a/d. is that something you really need to be worried about?


gregoire
del
ubk

[Donny]

Hi

Thank you for the comments. In response to the calibration issue my system is setup correctly as i periodically check this on a regular basis i.e. -15dbVu = 0dbFS. I did mention that on latter trials the difference became more audible. On early trials the re-recorded wave looked suspect especially on some transient information. This was the case with both the lavry Blues and 192's but...suffice it to say the audible difference was less noticable. However, a typical project that uses and outboard summer and various outboard pieces will go through a couple of passes of AD/DA. Further, at the mastering stage it will pass through a 3rd pass. I was merely interested as to how much this AD/DA will be affecting the sound. Im not at all an advocate of ITB and know that there is no 'real' resolution to what im implying above [at least in my case]. Yes its possible to stay all analogue but then this would be a very expensive means to an end.

[Tibbon]

To me, this SEEMS like it might be an issue with the A/D's and the D/A's not really matching up well, or being well suited for one another.

Let's say you take your 192, and over on the patchbay, you link the output of 1 to the input of 2, the input of 2 to the output of 3, etc... and then patch a signal generator into input 1, and let everything pass through ( I guess record enable everything in PT).

You'll see that the signal IS changing.

Now why is it changing? Is it because converters are inheriently evil things and play tricks with the signal? No.

It seems to me that (assuming that cabling is near perfect), that the levels being output by the D/A aren't EXACTLY the same as what's being 'seen' by the A/D just by a bit. Basically, even with the 24 bit dynamic range, that some information is falling 'between' steps consistently, or drifting off due to analogue circuitry. When it hits the next A/D it has to quantize the signal to a step, and somewhere in doing that several times, it's rounding off information incorrectly. I thought that anti-aliasing filters would fix a lot of this, but I suppose not.

It would in all seriousness, be easier to test this out, by using something like 2 or 4 bit converters, and using a square wave that is very close to the 'step point' bitwise in the convertor. You could easily tell if 'rounding' was occuring.

I might be 100% off, as I haven't have my caffinne yet this morning, so take that as you will.

I personally reconvert my things as needed, and could care less, as long as I make it sound good in the end.