Is this a valid way to test converters?

[KillerV]

I want to see what my converters do to sound that goes through it, so I took some recordings that I know fairly well and played them through my DAC's back into my ADC's, recording the results. I then A/B'd the original tracks to the ones I recorded through my converters, making sure the levels are the same.

Is this a valid way to test converters? If not, what methods can I use to learn a little more?

[Recording David]

For the purest test, why not just line up the 3 converters and test them playing back in real time at the same levels, one after the other?

If you can't do this for whatever reason, your test is fine, although you will also factor in the 'sound' of your ADC.

[sound_forward]

no.

best test is to compare analog source direct and the same source thru AD-DA conversion.

AD-DA conversion do much more damage to analog source (tape, vinly, analog synth...) then digital source (in your case, sound from your DA conversion).

And do not level match. I do AB test so that I first listen to one source, and start from silence and slowly add volume, and then switch to second source, also from silence to max volume.
Only then you will understand what your converters do to your source at differenet levels.

[Flying_Dutchman]

be careful with the gain settings.
If your converter is build to -22dbfs is +4dbu, than it is a "common practise" to level signals at -22dbfs with their rms level. I know, that some like to record hotter. But on some ADs: Hotter signals will push the analog stage before the converter harder, this could result in a less pleasing performance, more current, etc. ...
If you take a commercial cd release, peak is sometimes at -0,1dbfs and rms -6dbfs.
If you now take a loop for your soundcard and send the track out, it can sound different than recording 24 channels at -22bdfs rms recording.
Another problem could be the resistance of your ins and outs. Normally its a common practise to get this relation: out 1/ in 5 or more, for example microphone 200ohm, preamp 1kohm or more. So look at your soundcards data sheet for this.
Happy testing!

[Ethan Winer]

Originally Posted by KillerV

I then A/B'd the original tracks to the ones I recorded through my converters, making sure the levels are the same.

Even better, record ten passes through the converters to magnify the degradation. Most good converters sound the same after only one pass, but ten (or more) passes will really show up any differences.

--Ethan

[KillerV]

Originally Posted by Ethan Winer

Even better, record ten passes through the converters to magnify the degradation. Most good converters sound the same after only one pass, but ten (or more) passes will really show up any differences.

--Ethan

I've been wondering about this because what I recorded sounded seriously near identical to the original. I *think* I could hear a minute difference but it was really, really small. Now the thing is that I'm only using the converters once (on the way in from the source when recording), and then the only other time I can think of would be if I wanted to use some outboard gear or something like that where I would need to take the signal out of the DAC and back in again. That means I'll only do it twice at the most.

I also tried A/B-ing the analog signal with the converted one (mic'd voice, etc.), again with the same results - really small differences.

I just want to learn a bit more about this stuff after hearing lots of people make quite a big deal about converters.

[theblue1]

Originally Posted by sound_forward

no.

best test is to compare analog source direct and the same source thru AD-DA conversion.

AD-DA conversion do much more damage to analog source (tape, vinly, analog synth...) then digital source (in your case, sound from your DA conversion).

And do not level match. I do AB test so that I first listen to one source, and start from silence and slowly add volume, and then switch to second source, also from silence to max volume.
Only then you will understand what your converters do to your source at differenet levels.

This is not good advice.

First, the amount of 'damage' (distortion/inaccuracy) may be more apparent with one signal source or another, but it does not follow that AD/DA will "do much more damage to analog source" than to digital.

Second, using variable listening levels means your ears will have lost any sort of reference point. (Fletcher-Munson)

Making sure the sounds one is comparing have the same playback SPL in the monitor environment is one way of helping to assure that the ear has a relatively flat playing field in which to make its subjective judgments.

But bias error is a huge problem in such subjective judgment. For that reason, an ABX methodology is really called for, since, unless your converters are complete crap, the difference between the source and the round trip AD/DA version will likely be fairly subtle. And as a century of research into human perception has taught us, subtle variation in test material is the devil's playground when it comes to gauging subjective preference.

[Jaques Beraques]

Originally Posted by Ethan Winer

Even better, record ten passes through the converters to magnify the degradation. Most good converters sound the same after only one pass, but ten (or more) passes will really show up any differences.

--Ethan

A friend of mine did 100passes with an apogee.
The results were shocking!!

[Ethan Winer]

Originally Posted by Jaques Beraques

A friend of mine did 100passes with an apogee.
The results were shocking!!

I once did 20 passes with a $25 SoundBlaster card:

Original
After 1 pass
After 5 passes
After 10 passes
After 20 passes

--Ethan

[theblue1]

Originally Posted by Ethan Winer

I once did 20 passes with a $25 SoundBlaster card:

Original
After 1 pass
After 5 passes
After 10 passes
After 20 passes

--Ethan

This is a nice rubber-meets-road illustration of what you lose and what you don't going through a cheap (but not necessarily worst-case) soundcard.

While I've listened to these before, I've never run an ABX. While what my old ears can hear may not be of that much relevance to what the whippershappers can hear, I suspect that some of the differences I think are identifiable will become blurred as soon as I don't actually know what my source is. (Kudos to the folks at Foobar for building an ABX facility into it.)

[Jaques Beraques]

Originally Posted by Ethan Winer

I once did 20 passes with a $25 SoundBlaster card:

Original
After 1 pass
After 5 passes
After 10 passes
After 20 passes

--Ethan

Ethan,somehow I knew that you did!

[Flying_Dutchman]

Why should i run a signal 20 times through ADDA processings, to get informations about a converters behaviour?
In reality I record one time, thats AD x1.
Than i mix analogue, 1x DA and go back to DAW, 1X AD.

[Ethan Winer]

Originally Posted by Flying_Dutchman

Why should i run a signal 20 times through ADDA processings, to get informations about a converters behaviour?

Excellent question. These days even inexpensive converters are audibly transparent after one pass. So if you want to hear - versus measure - what sort of degradation occurs, you need multiple generations to make the degradation bad enough to hear.

--Ethan

[KillerV]

I actually wanted to ask the same question as Flying_Dutchman. So let's take this discussion a little further: if even inexpensive converters are still pretty good, what benefit would there be in buying top-of-the-line converters that folks are always telling me I need?

I don't mean to sound like an ass or an idiot about this - I really would like to know. After a friend and I watched "Shane Wilson's Guide To Mixing" he said that he wished he could do the parallell processing with the stems from ProTools but that he wouldn't try it with the crappy converters in his ProTools HD rig.

[getarzan]

Originally Posted by Ethan Winer

Even better, record ten passes through the converters to magnify the degradation. Most good converters sound the same after only one pass, but ten (or more) passes will really show up any differences.

--Ethan

Thanks Ethan.

This is the best idea I've heard yet , as a way to tell what kind of effect your converters actually have on your audio. I've been curious as H-LL to know how mine, ( 8 years old ) compare to converters developed recently. It's difficult to compare different units without having access to them , and who wants to buy something just because someone else "said" it sounds better than converters you are already using.

If the low end of your recorded source gradually "disappears" and/or there tends to be an increase in a particular frequency range in the mids or highs , at least you can be aware of it.

It sounds similar to the effects of several bounced tracks in analog recording of years ago , when recording engineers had to deal with the degrading highs after several stacked passes. They "had" to deal with it , and they did.

You may have just saved me some money!!!

[Ethan Winer]

Originally Posted by KillerV

if even inexpensive converters are still pretty good, what benefit would there be in buying top-of-the-line converters that folks are always telling me I need? I don't mean to sound like an ass or an idiot about this

I make the same point all the time, and people call me an ass and an idiot. So I'll take the heat for you.

Here's a test I did with a $25 SoundBlaster card:

Original
After 1 pass
After 5 passes
After 10 passes
After 20 passes

--Ethan

[Doc Mixwell]

Originally Posted by KillerV

Is this a valid way to test converters? If not, what methods can I use to learn a little more?

All the guys here on GS will tell you the loop test is the only way!!!! I intern suggest that if you're not going to be using your converters in this way to record and mix something, than it does not present a valid test. I would say its useless other than to kill time that you could be making music and listening to how your converters react/portray the audio you are sending through them. The only valid test, is compare and contrast, with all known variables accounted for with A/B/X. When you start WORKING with REAL WORLD APPLICATIONS, not 20/30/40 AD/DA pass through tests, I think you will gain a much better understanding of what these converter products offer the end user.

[KillerV]

Originally Posted by Ethan Winer

I make the same point all the time, and people call me an ass and an idiot. So I'll take the heat for you.

Here's a test I did with a $25 SoundBlaster card:

Original
After 1 pass
After 5 passes
After 10 passes
After 20 passes

--Ethan

Thanks man, I actually had my questions because I listened to that test you did. It came out way differently to what I expected it would be - and with a SoundBlaster card at that.

[Jules]

If possible try to enlist a friend or two to join you on the evaluations..

Then you can trade descriptions, become forced to articulate opinions, defend them, change your mind, validate decisions- get two brains in on the process..

I used to put my trainee engineers under pressure to describe the differences between various items gear... I would turn the flame up under them pretty high to come up with answers! It was fun .. and it added an extra dimension to gear shoot outs.. It got us talking about the differences, looking for language to describe it.

4 ears are better than 1!

[Ethan Winer]

Originally Posted by Jules

Then you can trade descriptions, become forced to articulate opinions, defend them, change your mind, validate decisions- get two brains in on the process.

If you play a track and listen very carefully, then play it again, and once more again, I bet you'll hear new things the second and third times. This is the main problem with subjective tests - ears are not very reliable. If you did this to a group of people blind, and didn't tell them it was the same track but said the A/D/A was changed each time, $100 says you'll still get a stimulating discussion of the "differences" from one A/D/A to the next.

--Ethan

[Doc Mixwell]

Originally Posted by Ethan Winer

ears are not very reliable.

Hi Ethan,

Do you speak for yourself? In this regard, I don't think you can speak for me.

[Audiop]

Originally Posted by Roc Mixwell

Hi Ethan,

Do you speak for yourself? In this regard, I don't think you can't speak for me.

Adam, Ethan speaks for humanity and science, he is spot on which is appearant to anyone that has looked into this seriously.

Now come come cooome join the other side!!


/Peter

[getarzan]

Hey Ethan,

You were talking about recording multiple passes through a converter to " magnify " the effect the converter has on the audio. Wouldn't you need a "presumably" really good quality D/A converter to run the audio through before it gets sent back through the A/D converter? I don't have an external D/A converter for monitoring right now. I only have the one from my board , though most people say it's essential for good monitoring/mixing .

How then would you recommend recording 10 or 20 passes to reveal the weaknesses of my converters?

[Ethan Winer]

Originally Posted by getarzan

How then would you recommend recording 10 or 20 passes to reveal the weaknesses of my converters?

The simplest way is to use the same sound card / converter for both D/A playback and A/D recording. That's what I did when I recorded 20 passes through my SoundBlaster card. You are correct that this lumps A/D and D/A together, with no way to separate out which affects the sound more. But if you can get three or more passes with no audible degradation, then that validates both the A/D and the D/A as being perfectly fine for one recording / playback. I'll also mention that D/A conversion is very simple compared to A/D. So playback quality is less likely to be an issue than recording quality. A perfectly functional D/A converter can be as simple as 16 resistors followed by a single op-amp output buffer amplifier. Versus A/D conversion which is at least ten times more complicated.

When I did the SB loop test I simply played a Wave file out of the SB sound card, and patched the output back into the input and recorded the result. I also balanced the volume very carefully to avoid accumulated level changes. With each pass I changed playback to the last file recorded, then recorded a new file.

--Ethan

[Ethan Winer]

Originally Posted by Audiop

Adam, Ethan speaks for humanity and science, he is spot on which is appearant to anyone that has looked into this seriously.

Indeed. The final frontier in stamping out audio BS is getting people to understand that nobody's ears are even 1/10th as reliable as a common multi-meter.

Adam, if you're willing, the next time I visit you guys let's do some blind tests in your control room. If you can reliably identify this stuff blind every time, dinner and drinks are on me. The first test will see if you can hear phase shift in a 10-foot length of line level wire as you once claimed.

Deal?

--Ethan

[getarzan]

Ethan , here's another one. What recorded sound source would you recommend for this test?

[Ethan Winer]

Anything that is full-range and sounds good. When I see people sheepishly question the quality you their low-end gear my standard suggestion is to take a CD that sounds really good, and record that analog from a CD player through their sound card. If it still sounds really good, then there's your answer.

Another good test is a sound source or sample that has a lot of high frequency transients. That's a good test of A/D conversion especially. I've used tambourines, triangles, claves, and the like.

--Ethan

[zacheus83]

I guess the question some of us are asking is:

If good converters don't degrade the sound and bad converters do (but only after multiple passes) then isn't testing the converters more irrelevant than trying to find a way to improve the sound BEFORE and AFTER the conversion??

For instance, if a crappy Mbox or M-audio product converter degrades the signal of 8-10 vocal tracks recorded into Pro Tools, and NOBODY can hear the degradation (according to Ethan) because no one (except the engineer and vocalist) were present during the recording, then why test the convertor in a way that is intentionally unintentional to the way we use them?

If I listen to convertor A and convertor B and can't hear an audible difference in 1 or 2 passes, and convertor B is $1000 cheaper - I'm buying convertor B and a nice set of monitors

People seem to forget that sound is meant to be HEARD. You get the idea.

[zacheus83]

Oh Yeah, and I RESPECT Ethan VERY MUCH! I've used his tips for placement of absorption in my room (along with modeCalc).

The last post was not an attack on Ethan, or anyone with an opposing point of view. Just a statement about my opinion when it comes to gear components contributing to the overall sound of a finished musical piece!

[getarzan]

I'm guessing that the accumulative effect of all your tracks together ; recorded through bad converters , will be audible yet hard to distinguish exactly WHY - precisely.

Ethan is not saying that bad converters don't degrade the sound ONLY after multiple passes. He is saying that because the effect of bad converters "might" sound slight during A/B testing , it's easier to find their weaknesses by amplifying the effect with multiple passes. Much in the same way that you might amplify different EQ frequencies to find problems , then adjust them appropriately after. That's my take. Ethan can stop by and bitch slap me , if I've got him all wrong.