D/A A/D Converter. Essential?

[reallyjake]

Hey, there's so many ways to "get there". And the strong link in the chain can definitely help the rest of it out.

Ethan gains miles and miles with his room.

I personally have my strong point in my preamps. That's where I can count on big gains. My converter is a lynxone and my home studio has no treatment. Actually, that's something I'm currently working on with great results. So yes, the poor link in the chain does pull you down, also.

So, it's different for everybody and everyone has a different setup to judge by. And this stuff is probably one reason why many of us have different opinions on what's really important.

Originally Posted by desotoslo

Wow... that is really interesting, Ethan.

After reading so much about the importance of high quality conversion on this board, your message is really a bit of a shock!

I am positive I heard a vast improvement in listening to the same music through an Apogee Duet over a MOTU Ultralite. That was of course just DA.

I am a dedicated headphone user (I know, that's already problematic, although I do use high quality stuff), but in that way I don't deal with room acoustics very much... and believe it or not, my mixes translate well to the outside world...


So in your expensive studio you use an Maudio card... wow.

Thanks for your advice and information, Ethan! It's great to get a different perspective.

[TurboJets]

Originally Posted by u b k

my experience with cheaper converters is that they are 'ringy' and 'peaky', and those unpleasantries very much do occur at the same freqs at all times. so this hash adds up quickly, and (e.g.) hardness at 4k and smear at 300hz and graininess at 10k give you a kind of soupy, blurry mess when the tracks start piling up, *especially* if you're eq'ing, compressing, and summing them digitally.

gregoire
del
ubk
.

This is exactly one of the points I tried to make earlier - touche'

Also, jitter is one of those anomalies that I used to hear from my crappy m-audio converters - it just drove me crazy. Jitter is especially audible to me (e.g.) in the sustain sections of ac guitar tracks. Since upgrading the converters of my m-audio box I don't hear that jitter any longer, thank god.

Also, lesser quality converters allow more "noise" into a signal than higher quality converters - that's scientific. Quality converters will allow like 4 bits (out of 24 bits) of noise into the signal whereas lesser quality converters will allow 7-8 bits (out of 24 bits) of noise into the signal - some products are even worse than 7-8 bits of noise.

All of those little differences add up to audible signal "distortion". When you reduce those factors of distortion by upgrading to higher quality converters you experience greater clarity and definition in your captures. It's elementary.

[notyetfamous]

For about 2 years I was running a ProTools HD rig at my house and at the studio with "stock" converters. The Digidesign 96 I/Os to clarify.

A few years back we made the change to Apogee A/D16 and D/A16
converters and immediately, the first session, noticed a world of difference between the two.

Now when I listen to sessions done before our upgrade, I almost always cringe when I hear the mix!

Of course there are variables from session to session, and It might not be a "scientific" comparison but if you try some nice A/D/A for at least a few sessions from tracking to mix, I can almost guarantee, you wont ever look back!!! Everything becomes more accurate. Transients were the most noticeable during tracking and a significant change in the responsiveness of the low mid range frequencies in most sources was obvious. During mixdown, everything seemed to have more space to breath (headroom) and my stereo image became a more obtuse angle giving me, once again more resolution in my L to R panning! Imagine having a bigger "box" to mix in.. If you can't mixdown analog then at least more accurate mixing in the box is worth the extra dough to me.


FYI...
After our conversion upgrade, when I thought it couldn't get better, we bought an Antelope ISOCHRONE Master Clock just to see what the difference was between that and the internal C777 clock on the Apogees. The Antelope did not get returned... Just another example of how your whole signal chain affects the final product... Bottom line is you are only as strong as your weakest link..

[Ethan Winer]

Originally Posted by jslevin

you divert the conversation into foolishness when you take the time to challenge a Mytek vs. 002 test. Mytek converters are well established and widely admired by thousands of engineers

I've never heard either of those converters, but I stand by the opinion that if any modern A/D/A sounds obviously colored it is most likely broken and not working up to spec. I will challenge any and all anecdotal reports that don't make sense to me. Lots of people say lots of things. Sometimes they're right, sometimes they're not. I see subjective opinions that I know for a fact cannot be right so often that I've learned to pretty much ignore all such statements. Especially about the quality of audio gear.

By coincidence, just this morning a friend sent me a pertinent quote:

"If it's consensus, it isn't science. If it's science, it isn't consensus." --Michael Crichton

Here's another quote that fits perfectly with this discussion:

"If 50 million people say a foolish thing, it's still a foolish thing." --Anatole France (1844-1924)

Yes, if we're going to have listening tests, they should be perfectly set up or else they're worthless.

This is exactly my point! Of those thousand anti-002 opinions you refer to, how many were the result of a truly proper test?

Well, then I guess I wish you great success and happiness with your SM57 and Antares Mic Modeler.

I would never argue that an SM57 could be coerced to sound like a U47, for all the reasons I listed in great detail above. Had I not explained all the limitations about EQ'ing one mic to sound like another I'd understand your comment. But I went out of my way to list many reasons that won't work, even if the basic response can be made similar. So I don't understand your sarcasm.

Again, I'm not here to pick fights, or be ornery just to be ornery. I have much better things to do. But there is a huge amount of misunderstanding out there about the science of audio, and plenty of open hostility to science too. That is what I respond to. The following sums up my position on A/D/A converters, and mic pres, and the joys of tube gear and analog tape, and everything else we see people argue about endlessly:

If it can be heard it can be measured. And if it can be measured it can be understood.

My sole interest is in understanding, and having others understand too.

--Ethan

[TurboJets]

Originally Posted by Ethan Winer

If it can be heard it can be measured. And if it can be measured it can be understood.My sole interest is in understanding, and having others understand too.
--Ethan

Then what do you do when it comes to women Ethan? Because sometimes I think I'll never understand women.

But, there's a quote fitting to that question from Oscar Wilde - "Women are meant to be loved, not to be understood."

Sometimes observation will suffice until science catches up. The power of observation is just as important to recognize.

[desotoslo]

easy for oscar wilde to say... his sexual drive didn't depend on it.

[Ethan Winer]

Heh, fortunately audio gear is nothing like women.

I'm flying out the door now, but I think we should continue this tomorrow and talk about what makes for a valid comparison test, and what doesn't, and why. Any interest?

--Ethan

[kedbear]

Originally Posted by Ethan Winer

Heh, fortunately audio gear is nothing like women.

I'm flying out the door now, but I think we should continue this tomorrow and talk about what makes for a valid comparison test, and what doesn't, and why. Any interest?

--Ethan

That would be great. Even better though, if you could get hold of a Mytek or equivalent and show what you mean by a scientific test and post it's outcome for all to see.

[Dean7]

Originally Posted by kedbear

That would be great. Even better though, if you could get hold of a Mytek or equivalent and show what you mean by a scientific test and post it's outcome for all to see.

Good idea. Let's all put $5 in to a hat to get Ethan enough cash to buy some Mytek converters. :D

-Dean

[TurboJets]

In my studio my own observations are the only thing that matters, and my observations are as "valid" (thank you) as it gets. I hate to say this Ethan but it's rather arrogant to offer to define what's "valid" and what's not. Reminds me a little of the "I'm The Decider" moment.

[Dean7]

Originally Posted by TurboJets

In my studio my own observations are the only thing that matters, and my observations are as "valid" (thank you) as it gets. I hate to say this Ethan but it's rather arrogant to offer to define what's "valid" and what's not. Reminds me a little of the "I'm The Decider" moment.

Well, your poit of view makes sense to me. But, you're not saying that anything is "better" than anything else. You're just making the (valid) point that if you think it's better that's all that matters to you.

[kedbear]

Originally Posted by TurboJets

In my studio my own observations are the only thing that matters, and my observations are as "valid" (thank you) as it gets. I hate to say this Ethan but it's rather arrogant to offer to define what's "valid" and what's not. Reminds me a little of the "I'm The Decider" moment.

More to the point though, this is an opportunity for Ethan to demonstrate what he is talking about. How many scientific tests have actually been posted here? It would seriously determine the BS from fact. I would love to hear two converters, both that have similar spec's but vary wildly in price put next to each other. Or another classic, a mackie preamp vs expensive one, through a good converter, with a mic split, through a good room, as Ethan was talking about.

Do you think it would be possible to set something up Ethan? To demonstrate similar specs that vary wildly in price next to each other?

[Ethan Winer]

Originally Posted by kedbear

That would be great. Even better though, if you could get hold of a Mytek or equivalent and show what you mean by a scientific test and post it's outcome for all to see.

Originally Posted by Dean7

Good idea. Let's all put $5 in to a hat to get Ethan enough cash to buy some Mytek converters.

Well, we don't need physical gear present to discuss what constitutes a valid test.

First, let's look at some things that are not valid:

Comparing two different performances is never valid because nobody can sing or play exactly the same way twice. The tiniest bit of sibilance can make one vocal sound brighter or clearer than another. And nobody can stand or sit in exactly the same place. If the sound source or microphone move even 1/4 inch, the response at high frequencies can change a large amount.

Likewise, every session is different. Different performances (as above), and also slightly different microphone placements. When someone says they replaced [insert gear type] and heard an obvious difference at the next session, there are just too many variables at play to conclude anything. This doesn't mean the new device is not better! But it's not a valid test.

When considering raw fidelity only - not color which is subjective - the best way to assess the performance of audio gear is to simply measure it. Subjective descriptions like "fuller" and "brighter" directly relate to frequency response. If you perceive a new mic pre to be fuller than another, this is easily confirmed by a simple response measurement. Clarity can be a flatter (or exaggerated) HF response, but it can also be lower distortion. Distortion is also very easy to measure.

If the response of an audio device is within 0.1 dB from 20 Hz to 20 KHz, and the sum of all noise and distortion is 90 dB below the signal, that device can be said to be audibly transparent. Any two devices with specs that good or better will indeed sound identical. But we know there are nay-sayers who claim to hear differences in gear even when both measure well enough to be considered transparent. This must then be resolved with a proper listening test. This will reveal whether a difference really can be identified, or if it's just the listener's imagination and expectation.

To be valid, a listening test must be done blind where the listener does not know which device is playing at the time. Even better is a double blind test, where both the listener and the test person switching devices do not know which is which. This avoids the test person from inadvertently revealing which is which with facial expressions or gestures or other such clues.

Besides the need for blind testing, another requirement is that both devices be level matched to within 0.1 dB. If the volume from one source to another varies even a little, the louder version will usually be preferred as sounding better. And the softer one will seem to have less clarity and presence even if the frequency responses are identical. This is Fletcher-Munson at work.

There's much more to this! But it's a good starting point for a discussion. If you look through the numerous posts here on Gearslutz that report on a "shootout" of gear, use a critical eye to see how many of these rules are violated thus rendering the shootout useless.

--Ethan

[Ethan Winer]

Originally Posted by kedbear

How many scientific tests have actually been posted here? It would seriously determine the BS from fact.

Separating BS from fact is the point. There's no reason TurboJets should not use whatever criteria he wants to decide what gear to buy and keep. His opinion is all that matters to him! And there's nothing wrong with that.

But most people don't stop there, and go on to use their own subjective opinions to advise others how to spend their hard earned money. I like to think we all can agree that The Truth (tm) is worth pursuing. And that's all I'm trying to do.

--Ethan

[jslevin]

Originally Posted by Ethan Winer

"If 50 million people say a foolish thing, it's still a foolish thing." --Anatole France (1844-1924)

Ethan, it's a cute quote, but it's also a diversion — although I agree that if you can get 50 million people to say there's no material differnece between a 002 and a Mytek, then it would still be foolish.

I don't need a scientific test to tell me Ella Fitzgerald sang better than William Hung, and by the same token lots of engineers know which tools are helping them to do high-quality work. I fundamentally don't even believe in listening tests, I believe in experience working with a tool.

Get yourself a 002 and a Mytek converter, measure nothing, use them both for yourself, then come back here and tell me you didn't find the difference to be obvious.

JSL

[Dean7]

Originally Posted by Ethan Winer

Separating BS from fact is the point. There's no reason TurboJets should not use whatever criteria he wants to decide what gear to buy and keep. His opinion is all that matters to him! And there's nothing wrong with that.

But most people don't stop there, and go on to use their own subjective opinions to advise others how to spend their hard earned money. I like to think we all can agree that The Truth (tm) is worth pursuing. And that's all I'm trying to do.

--Ethan

So, Ethan, do you know of any software that would do a good diff for WAV files? I was thinking about the issue of conversion (and how much of a difference it makes), and it's an easy issue to put to rest. All you need to do is generate a waveform in software, output said waveform from your converter back in to your converter (it would have to be ad/da or you'd have to use some sort of reference grad converter for the da), then use a diffing tool to compare it to the original.

What do you think about that testing methodology? Is there something I'm missing here that makes that a bad test?

Thanks,

-Dean

[allencollins]

Originally Posted by Ethan Winer

Subjective descriptions like "fuller" and "brighter" directly relate to frequency response

They directly relate to the 'timbre' of the sound. I would have figured your for a student of psychoacoustics? no?

Anyway do a recording with your $200 A/D and then do one with a Mytek 8x192
I bet 90% of the folks here COULD tell the difference despite what you think.

I think most folks can tell a cheapo converters within a couple seconds.
You should really rent a UA2192 or a radar system for a weekend do some recording and get back to us.
I will guarantee you would change you mind about a/d.

In fact your opinions on a/d make me real skeptical about..........

[Dean7]

Originally Posted by allencollins

They directly relate to the 'timbre' of the sound. I would have figured your for a student of psychoacoustics? no?

Anyway do a recording with your $200 A/D and then do one with a Mytek 8x192
I bet 90% of the folks here COULD tell the difference despite what you think.

I think most folks can tell a cheapo converters within a couple seconds.
You should really rent a UA2192 or a radar system for a weekend do some recording and get back to us.
I will guarantee you would change you mind about a/d.

In fact your opinions on a/d make me real skeptical about..........

I would really be interested to hear the results of a test like that. The only problem with most people being able to hear cheap converters is that they are usually coupled with cheap pres and / or mics (because people with MBox 2s usually have them because they don't have the cash to buy something else). So, I'd be really interested to hear a scientific test.

Sure, if you listen to the GS people that record with Mytek converters they are going to sound way better than people with more humble converters, but people with Mytek converters generally will have good rooms and pres as well. So, it's hard to tell what's what. Know what I mean?

-Dean

[Ethan Winer]

Originally Posted by Dean7

So, Ethan, do you know of any software that would do a good diff for WAV files?

I've done that by loading two files into separate tracks in Sonar, and flipping the polarity on one track. This is called a null test, and it reveals the differences between two files and at what level those differences occur.

There's one potential problem with a null test. Sometimes phase shift in one file changes the waveform so it can't null with the other file, even though the frequency response and distortion etc are otherwise identical. Phase shift alone is benign and inaudible, so that part is not a problem. But it still prevents the nulling that should otherwise confirm that the two files are the same or very close.

There's also the ABX software meant for this specifically:

www.pcabx.com

(The above site seems to be down at the moment.)

I'd be really interested to hear a scientific test.

Right, this is the key. Also, I don't know why some people here are so defensive about this device or that. I never said all A/D/A converters are identical. My two main points have simply been:

1) You really don't know until you do a proper controlled test.

2) Pro-quality A/D/A converters probably vary less than some people think.

--Ethan

[Dean7]

Originally Posted by Ethan Winer

I've done that by loading two files into separate tracks in Sonar, and flipping the polarity on one track. This is called a null test, and it reveals the differences between two files and at what level those differences occur.

There's one potential problem with a null test. Sometimes phase shift in one file changes the waveform so it can't null with the other file, even though the frequency response and distortion etc are otherwise identical. Phase shift alone is benign and inaudible, so that part is not a problem. But it still prevents the nulling that should otherwise confirm that the two files are the same or very close.

There's also the ABX software meant for this specifically:

www.pcabx.com

(The above site seems to be down at the moment.)

So when you say you loaded 2 tracks in Sonar you mean you had the "original" track and then sent the "modified" track out through your D/A conversion back in through your A/D conversion then flipped polarity? What did the results yield? Were the files nearly identical?

Anyway, I think some sort of diffing tool would be great for this. In terms of mathematics it actually wouldn't be incredibly hard if you had the files in PCM WAV... you could invert phase, then for each sample add the amplitude of track 1 to track 2 (or subtract it depending on how it's stored) and they SHOULD come up with a 0 (or whatever number represents the middle of the waveform). If anything comes up plus or minus you'd be able to create a report on it or something.

But, I could be wrong. There are still a lot of things about digital audio that I don't understand (or I understand but not well enough to think of all of the problems with my logic).

Originally Posted by Ethan Winer

Right, this is the key. Also, I don't know why some people here are so defensive about this device or that. I never said all A/D/A converters are identical. My two main points have simply been:

1) You really don't know until you do a proper controlled test.

2) Pro-quality A/D/A converters probably vary less than some people think.

--Ethan

Yeah... I REALLY hope there's a difference between Mytek converters and an MBox Pro, but so far my tests where I've been able to isolate JUST conversion have yielded identical results (I've only been able to do an extremely limited test between a ProFire 2626 and a Lucid converter thus far, so by no means do I have conclusive evidence). I write software for a living, and sometimes hearing people talk about software is shocking because they've built up an entire mythology around why things behave the way they do (supported by some "evidence" usually, and sometimes a somewhat logical conclusion to arrive at) that has nothing to do with reality. So, I bet people who build AD/DA converters for a living are sometimes shocked by the mythology people have built up around them and are left scratching their heads.

-Dean

[Ethan Winer]

Originally Posted by Dean7

So when you say you loaded 2 tracks in Sonar you mean you had the "original" track and then sent the "modified" track out through your D/A conversion back in through your A/D conversion then flipped polarity? What did the results yield? Were the files nearly identical?

I was just describing a generic technique for comparing two files that are supposed to be the same, or at least similar. For example, this is a good way to see how audible the effects of dither are. You export a DAW mix at 16 bits with no dither, and then another with. Or export two files with different dither types. Then when you load and null the files, only the differences remain. If the differences are small you can crank the speakers to hear them. Or export that to a new Wave file for analysis.

I write software for a living

Cool! I did that too for many years, back in the late 1980s and early 1990s when DOS ruled. I had a company that sold programmer tools to other programmers. Mostly low-level subroutines written in assembly language for speed, that other programmers didn't know how to write or would take them months. I did pretty well with that, and also wrote DOS utilities and contributed to the Languages column at PC Magazine. What sort of software do you write?

--Ethan

[Dean7]

Originally Posted by Ethan Winer

I was just describing a generic technique for comparing two files that are supposed to be the same, or at least similar. For example, this is a good way to see how audible the effects of dither are. You export a DAW mix at 16 bits with no dither, and then another with. Or export two files with different dither types. Then when you load and null the files, only the differences remain. If the differences are small you can crank the speakers to hear them. Or export that to a new Wave file for analysis.

Gotcha. Yeah, that's a good testing method. BTW, since you mentioned dither have you read Massey's blog entry on dither? If not, it's a great read: smassey.com. :)

Originally Posted by Ethan Winer

Cool! I did that too for many years, back in the late 1980s and early 1990s when DOS ruled. I had a company that sold programmer tools to other programmers. Mostly low-level subroutines written in assembly language for speed, that other programmers didn't know how to write or would take them months. I did pretty well with that, and also wrote DOS utilities and contributed to the Languages column at PC Magazine.

Somehow that doesn't surprise me.

Originally Posted by Ethan Winer

What sort of software do you write?

I work for a Microsoft partner shop doing primarily ASP.Net, Winforms, etc. in C# (probably a far cry from your low-level subroutines!), so I mostly write business software. Our usual stuff ranges from billing systems to eCommerce and document management systems. It can be incredibly fun, but it's not incredibly challenging, so I've also been writing some lower-level stuff lately just for the heck of it (last night I finished version 1 of a VST plug-in, for example) because the things that interest me are the problems that seem to be worth solving.

-Dean