How Many Think Mixing "In the Box" Doesn't Sound Right?

[Anderton]

>

 

I think the truth is revealed by the "golden ears" types who participate in the "external summing vs. ITB" tests -- they have no problem agreeing that they often sound different. But they have a hard time accurately identifying which is which in double-blind tests, and an even harder time agreeing on which one sounds better.

 

Auditory memory is horrible when it comes to volume levels and other characteristics of hearing. You can remember the name of a classmate in 3rd grade, but auditory memory fades rapidly. As a result people often compare sounds to sounds stored in memory, which are far from perfect. This is why I wish we'd see more A-B testing with correct testing protocols. The results can sometimes be surprising, like when top recording engineers choose a low-end solid-state preamp over a big-bucks tube preamp when they can't see the logos

 

OTOH one of my favorite "golden ears" stories was when CBS was trying to push its disastrously inept and stupid "copycode" technology, which put a notch at somewhere around 3 kHz (I forget exactly) in CDs in order to prevent people from making cassette copies (I'm not making this up). A bunch of creme de la creme engineers were assembled, and said they couldn't hear a difference, and gave the technology their blessing.

 

Then the National Bureau of Standards was summoned by Congress to do some tests on the copycode thing before they passed a law requiring it. So they grabbed a bunch of random people off the street, and an overwhelmingly high percentage of people (like 85% of something, I don't recall) could pick out the copycode version, every time, in double-blind A-B tests.

 

My point? I dunno...just random musings. The last CD I mastered was all-digital, recorded in Pro Tools, mastered in digital. It sounds effing great and picked up an award. Because it was digital? No. Because the artists, studio acoustics, mics, and preamps were all simply fabulous -- as good as it gets. They were recorded faithfully, and that's why it sounded so good. I seriously doubt it would have sounded any better going through an analog console. Although it might have sounded different

[Anderton]

>

 

Well there are quite a few people who think that all our problems with music started with the adoption of even-tempered tuning. Before then, music was used a lot for healing; afterward, the concept of music as medicine started fading out in favor of the concept of music as entertainment.

[Lee Flier]

 

Originally posted by Anderton I think the truth is revealed by the "golden ears" types who participate in the "external summing vs. ITB" tests -- they have no problem agreeing that they often sound different. But they have a hard time accurately identifying which is which in double-blind tests, and an even harder time agreeing on which one sounds better.

 

 

I'm not sure which tests you're referring to... but I can tell the difference in most cases, and it seems very obvious to me and not something you'd have to be a golden-eared type to hear.

[Lee Flier]

Originally posted by Anderton Well there are quite a few people who think that all our problems with music started with the adoption of even-tempered tuning. Before then, music was used a lot for healing; afterward, the concept of music as medicine started fading out in favor of the concept of music as entertainment.

 

Interesting idea, but I think it's more to do with the fact that the rise of even tempered tuning more or less went right along with the rise of Western science and medicine. And once Westerners started "worshipping" science they began to doubt the power of music as a healing force, even though doctors back then pretty much didn't know jack.

 

Not that I'm necessarily arguing the point about equal temperament, it may very well be screwing things up too.

[Anderton]

>

 

I indeed said they can agree there's a difference. What's harder is for people to unambiguously identify which is which, then make a value judgement about which is "better."

 

For example, some people who can identify the difference might feel digital is clearer, or that analog is wamer -- both difficult qualities to quantify, although I think there's some evidence that people associate "stuff with good audio transformers in the signal path" as "warmer," which I would link to the midrange ringing/overhang. And I think the "clarity" of digital might be due to distortion creating additional harmonics...hey, who knows.

 

I still think DSD sounds bitchin' and can't understand why there isn't more interest in that technology. It sounds like digital without the "harsh" and analog without the "fuzzy."

[Lee Flier]

Originally posted by Anderton I indeed said they can agree there's a difference. What's harder is for people to unambiguously identify which is which, then make a value judgement about which is "better."

 

OK I guess I wasn't clear enough: I can tell in most cases which is which, and also that the console mix sounds obviously "better."

 

For example, some people who can identify the difference might feel digital is clearer, or that analog is wamer -- both difficult qualities to quantify,

 

See, to me those aren't even the right terms to describe the differences that I hear.

 

What I hear is what Zeronyne described - in a digital mix the soundstage closes up. It doesn't have as much depth or breadth - also difficult to quantify, but that's what it sounds like. And verb tails, as I said, get swallowed. I'm not sure what causes this but I suspect it's lots of tiny timing/phasing problems added together. I don't claim to be an expert about the underlying cause, but I just know what I hear.

 

Also, the extreme highs and lows get compromised. The highs get brittle and/or swallowed up, and the lows get a bit "tubby" sounding and/or also swallowed up, compared to a console. I've also noticed that the low end and high end thing happens to an extent when doing a digital recording vs. on tape, even given really good converters. The soundstage problem seems to go away though when mixing with a console, regardless whether the tracks are analog or digital.

 

I still think DSD sounds bitchin' and can't understand why there isn't more interest in that technology.

 

My memory is a little fuzzy because it's been a little while since I looked into it, but I thought the reason it isn't being used for multitrack recording is because you can't edit it without converting it to PCM first, which kinda defeats the whole purpose. So there is interest in it as a distribution and archiving medium (and I guess lately not even that, which is a shame), but not multitrack. Last I heard somebody was working on some 8 bit "super DSD" format which could be edited... but again that was awhile ago. Anybody's welcome to correct me if I'm wrong and/or fill me in on what I've missed in the last couple of years.

[Anderton]

>

 

I'm pretty sure there's a definite correlation between clock accuracy/jitter and soundstage. That makes sense to me; if the data isn't being parsed out precisely to two channels, then you're going to get commonalities that cause collapsing.

 

I first noticed the "big soundstage" effect in Ensoniq's PARIS system. I asked the engineers there why their system had a much better soundstage than what was out there at the time, and they said it was because they knew analog engineering as well as digital. They seemed to feel that the analog circuitry and board layout was really crucial to get a good digital sound.

[Lee Flier]

Yeah I think you're right Craig, about clock accuracy, but that only comes into play at the converters. I think there are further timing issues that happen during mixing - I think there is still latency happening even with supposed latency correction, and that automation/gain changes done digitally degrade the signal more than we realize, and maybe other stuff too.

 

I was going to say though, that the PARIS system didn't seem to have that problem! At least, far less so than any other digital board I've ever heard. I wonder why that is, and I also think it sucks that they are no longer being made.

[Anderton]

>

 

YES! As I understand it, because of digital's extended/phase accurate low frequency response, those volume changes are actually adding subsonic modulation signals. You know how many people note how much better digital mixes sound when you roll of the low end? I think that's part of it. You never had this problem with analog because a) DC never made it into the signal path (or at least it wans't supposed to), and b) if it did, there were caps to get rid of it -- although those caps also caused low end phase shifts you DON'T get with digital. For dance mix bass, I'll take an analog synth recorded digitally over an analog synth recorded via analog any time, precisely for that reason.

 

I've done digital mixes with "mystery low end" problems and upon further analysis, many of them related to subsonic signals. Once those were gone, the low end sounded a zillion times better and come to think of it, the whole mix just opened up a whole lot more. These days, as a matter of course, I add pretty gentle highpass filtering because thanks to the dB/octave rolloff, the octaves get bunched together at the low end and by the time you hit something like 0.05Hz, the response is way down.

 

Note that the problem isn't so much the subsonics per se, but how they affect the parts you CAN hear.

[Lee Flier]

Yeah that makes total sense Craig, and I too have a tendency to high-pass everything in digital mixes, to get rid of the subsonics. It does open up the mix a lot, but I still suspect there's more going on with gain changes, including edit crossfades.

 

(and just to love on the Kjaerhus EQ a little more - that's one of the best things about it is that it has a VERY nice high pass filter, which really does a great job of opening up a track, has very low CPU usage and you can use it across a bunch of tracks without it starting to degrade the mix in its own right).

[MitchG]

 

Originally posted by Lee Flier I think Lynn's test is a great idea, but as I've said all along since its inception (and I know Lynn agrees so this is not a diss on him at all), it's only the first step and no final conclusions can be drawn from it. Reason being that there was absolutely no processing done on the tracks, not even gain changes. Start adding automation, plugins, and other stuff that is likely to happen in the real world and the difference becomes much more apparent. Still, the fact that there are differences in summing between the different DAW's at all is interesting.

 

 

To me the fact that there are differences with just pure summing is pretty critical. It seems that everything else builds on that...it better be "right"! Once that is "right," (whatever that is) then you can start adding processing, etc., on top.

[MitchG]

 

Originally posted by Lee Flier Yes I can nearly always tell the difference between an ITB mix and a console mix

 

 

I'm curious about this -- and first let me qualify by saying that I completely believe you.

 

But what does this mean? If I play you a CD of a song you've never heard before, you can tell if it was mixed in the box or through a console? Or you can tell in comparison to hearing other stuff? Have you tried double-blind tests? Ever tried, say, where you were just presented with 10 unidentified songs played consecutively in a row, some OTB, some ITB, and you selected which were done in the box, and which weren't?

 

Is there a point, such as mixing ITB, but using external analog mastering, or one piece of external processing gear on a track, or two pieces of external processing gear, that you can note the difference -- in other words, how far OTB does it have to be before you can't tell? (I'm thinking of the recent review in SOS, where the reviewer was looking at the new Neve summing box. He says that he got the same sonic results whether actually summing with the box, or just passing a stereo mix through it.)

 

(Hopefully you know me well enough to know I'm not challenging you, I truly am curious...these subjective/objective things intrigue me.)

 

One thing that was interesting to me about Lynn's disc (and I SWEAR I'm not trying to hype it) were the differences I heard between different consoles, both analog and digital. It made it nearly impossible to establish a reference (at least until you knew which mix was which, at which point the test is pretty much nullified) - you could only choose what you "liked" better; purely subjective.

I did get to where I could pick out certain consoles, etc. (I was doing double-blind listening, but I got to where I could say, "this track is the same as the one you played 3 times ago"), but it took a fair amount of listening, to my ears the differences were pretty subtle, even in a well-tuned listening environment. And the ones I picked as analog, weren't always the analog ones!

[harmonicat]

I've always been of the mind that I use technology to reinforce my own sense of traditionalism-i.e. Just because I use a Keyboard Workstation to lay orchestral tracks does NOT mean that I do "electronic music",not by ANY stretch.I'm not John Williams and I don't get to conduct an orchestra,so...

 

I think Analog and Digital should *both* be employed wherever possible because there are advantages to both,as stated many times by many people on many forums.Personally,I love the idea of a Personal Digital Studio that lets you do things that were practical science-fiction 16 years ago,at least $$$ speaking.But at the same time,I'm a child of vinyl and I like that warmth aspect too.Even in this Hi-tech time that we live in,there are still tools that we need( #2 shovel anyone?)-tools that still persist and aren't likely to go away anytime soon,in spite of all the persistant "it's the death of *whatEVER*" proclaimations.I suspect it's always going to be that way too.

[Lee Flier]

Originally posted by MitchG But what does this mean? If I play you a CD of a song you've never heard before, you can tell if it was mixed in the box or through a console? Or you can tell in comparison to hearing other stuff?

 

Usually I can tell without any other reference. Generally there will be something that jumps out at me as either "There is no way that's ITB; it has to be a console" or "That's gotta be ITB." Usually the giveaway is the reverb tails, or something at the extreme high or low end.

 

Have you tried double-blind tests? Ever tried, say, where you were just presented with 10 unidentified songs played consecutively in a row, some OTB, some ITB, and you selected which were done in the box, and which weren't?

 

Maybe not 10 in a row, never tried that, but yes I have done it with single songs on quite a few occasions. One of them in fact was Lynn's "Neve vs. Pro Tools" test where he mixed the same piece of music through a Neve console and a PHTD rig. The tracks were recorded digitally on a RADAR. He used no effects on either mix except an outboard reverb (the same one in both cases) and he matched the levels in the mix as closely as possible. I was able to identify which was which even in MP3 format on my crappy computer speakers.

 

I would say that I'm more reliable at being able to identify which mixes couldn't possibly have been done ITB than which ones were definitely ITB, but in general, you could play me either one and I can usually tell.

 

Is there a point, such as mixing ITB, but using external analog mastering, or one piece of external processing gear on a track, or two pieces of external processing gear, that you can note the difference -- in other words, how far OTB does it have to be before you can't tell?

 

If any processing has been done ITB before summing OTB, it makes it harder to tell - it sounds more ITB to me at that point. I would say that I can consistently hear the benefits mainly if all of the summing, gain changes/automation, and effects have been done on a console, otherwise there is some, but not as much, difference between that and full ITB.

 

(I'm thinking of the recent review in SOS, where the reviewer was looking at the new Neve summing box. He says that he got the same sonic results whether actually summing with the box, or just passing a stereo mix through it.)

 

I didn't see the review - but yeah that's exactly what I'm talking about and it doesn't surprise me. There may be something about the summing box that imparts a pleasing character of its own, which would happen whether individual tracks went through the box or the stereo mix. But the full benefits to the sound stage can't be realized unless each track's level setting and effects are done via the console.

 

One thing that was interesting to me about Lynn's disc (and I SWEAR I'm not trying to hype it) were the differences I heard between different consoles, both analog and digital. It made it nearly impossible to establish a reference (at least until you knew which mix was which, at which point the test is pretty much nullified) - you could only choose what you "liked" better; purely subjective.

 

Well yeah, in a way - I agree that all analog consoles sound different from each other and most digital summing engines sound different from each other. But there are still certain things that any midline or above analog console consistently does that I have never heard any digital console or mix in a box do. And there are also certain sonic characteristics (and not "good" ones IMO) that you will usually hear in an ITB mix that you won't hear in a console mix, unless you've already done a lot of processing ITB before going out to the console.

[d. gauss]

here's my wrench in the whole situation..

 

more often than not, ITB stuff has been recorded in the box as well.

 

who here has mixed stuff in the box that was originally tracked ALL analog then transfered to the box to mix? i have, and it was interesting.

 

-d. gauss

[Lee Flier]

Originally posted by d. gauss more often than not, ITB stuff has been recorded in the box as well. who here has mixed stuff in the box that was originally tracked ALL analog then transfered to the box to mix? i have, and it was interesting.

 

I've done that quite a bit... am in the middle of doing some of that now actually.

 

It helps a LOT. Especially with the low end coherency thing.

[MikeRivers]

Originally posted by Lee Flier I think Lynn's test is a great idea, but as I've said all along since its inception (and I know Lynn agrees so this is not a diss on him at all), it's only the first step and no final conclusions can be drawn from it. Reason being that there was absolutely no processing done on the tracks, not even gain changes. Start adding automation, plugins, and other stuff that is likely to happen in the real world and the difference becomes much more apparent.

But once you start doing that, you're no longer comparing the same thing. You might as well be comparing a Neve and API equalizer on the same track, and you'd expect a difference.

 

I think that the choice for most people is already made based on economics, whether it's a low budget home studio, a no limit major production, or a producer working for a major label who's trying to stay within budget (or keep a little more money for himself).

 

The purely summing demonstrations are indeed interesting, but I don't think any of them have every really proved anything conclusively. There was an interesting one in a magazine not too many months ago where they compared a ProTools straigh sum to a $4K Dangerous box, a Behringer mixer, and one or two in between. They took some care to match the gain for each - from channel input to left and right out, and found an insignificant difference (which of course means something differerent to everyone) between the stereo outputs.

 

Go figure. Too many people think they have a problem that buying a new piece of hardware will solve, and that's why we have so many gadgets and plug-ins on the market. A REAL problem is likely to be an inaccurate monitoring setup, a ground loop, or incorrect gain management. Or just garbage in. Maybe there really are a lot of studios where there's really something wroing, but I'm not used to that. I guess I don't get around enough.

[MikeRivers]

Originally posted by Anderton > But unfotunately, a lot of people haven't.

More significantly, there haven't been plenty of people who have been around learning it for 20 years. While the recording system, at any price point, has become technically better, skills don't improve quite as fast.

 

Fiddles have been around for 1000 years, and I still can't play one very good. But at least I have the good sense not to do it in public.

[MikeRivers]

Originally posted by Lee Flier Well sure, lots of people don't know how to do digital right. Lots can't do analog right, either. That's kind of beside the point.

Well, yes and no. In theory, it doesn't matter, but in practice it distorts any of the experiments and demonstrations that have been discussed or published. Engineers with 25+ years of experience, whether they were early or late (or not at all) adopters of digital mixing, learned on a fully analog system. If they were able to stay in business for that long, they must have learned to do something right. Today's ProTools "expert" may be a highly skilled computer operator, but his experience is in solving existing problems rather than preventing problems from occurring. Probably the best situation is when you have a team consisting of a good engineer and a good DAW operator - where each can do (only) what they do best.

The point being that even if you take two engineers who are both doing everything right in their respective formats, and are both great mix engineers, ITB is still not going to sound as good as a decent console.

How can you tell? They won't do the same thing on the same project, so there can never be an accurate comparison. Do you think that an experienced engineer is going to put his name on a "brittle" sounding mix or track unless he wanted it to sound that way? The most important thing that one gains from successful working experience is to recognize when there's a problem. If you are able to do this (and apparently you can) then it's up to you to do something about it, or alternatively, decide that it's not what you might do under other conditions, but it's OK.

[MikeRivers]

 

Originally posted by Anderton OTOH one of my favorite "golden ears" stories was when CBS was trying to push its disastrously inept and stupid "copycode" technology, which put a notch at somewhere around 3 kHz (I forget exactly) in CDs in order to prevent people from making cassette copies (I'm not making this up). A bunch of creme de la creme engineers were assembled, and said they couldn't hear a difference, and gave the technology their blessing. Then the National Bureau of Standards was summoned by Congress to do some tests on the copycode thing before they passed a law requiring it. So they grabbed a bunch of random people off the street, and an overwhelmingly high percentage of people (like 85% of something, I don't recall) could pick out the copycode version, every time , in double-blind A-B tests.

I participated in that NBS test. It wasn't exactly a bunch of people off the street, the group I was in was from members of our local AES chapter (and friends and associates) - not all certified golden ears, but people who had more than a passing interest in audio. Copycode wasn't to prevent people from making cassette copies, it was to prevent them from making DAT copies. I have the report here (76 pages plus a bit more than that in data in several appendices), and the notch frequency was 3840 Hz.

 

The session started with a hearing test to make sure that we could all hear in the range where the notch was placed. Then, there was a couple of hours of listening for training purposes. The group was divided in half, with half listening on headphones, the other half listening on speakers, and then we switched.

 

Finally, we had the actual scored listening. Sources were analog recordings, digital recordings, and synthesizers. I scored 95%, and everyone in my group scored significantly better than random choice would predict.

My point? I dunno...just random musings.

My point? This was a well-conducted test. I'm not sure that any ITB vs. console test has been so well conducted. Maybe if Congress gets involved in that debate, NIST (new name for NBS) will get involved and do it right. Interestingly, this was the first time that Congress got involved in making a decision that affects commercial application of a technology. Fortunately they solved it by commissioning a scientific test rather than getting free trips to famous golf cources.

 

The test concluded that:

 

1. The system does not achieve its purpose. It does prevent copying of notched material much of the time, howver, for about half the tracks studied, the system exhibited false negatives - notched material was able to be recorded. Also, the system exhibited false positives, preventing recording of un-notched material. This was the killer as far as Congress was concerned.

 

2. In double-blind listening tests, 87 subjects listened to a prerecorded tape of short selections from 24 sources - 20 CDs and four selections produced on a keyboard synth. Although the effect of the notch was fairly subtle on some musical selections, there are some selections for which the subjects easily detected differences between notched and un-notched material. I recall one where the oboe disappeared in the notched version.

 

3. They system can easily be bypassed.

 

NBSIR 88-3725 "Evaluation of a Copy Prevention Method for Digital Audio Tape Systems" - it's probably on the web somewhere.

The last CD I mastered was all-digital, recorded in Pro Tools, mastered in digital. It sounds effing great and picked up an award. Because it was digital? No. Because the artists, studio acoustics, mics, and preamps were all simply fabulous -- as good as it gets. They were recorded faithfully, and that's why it sounded so good.

And that's something that you probably didn't have the skills and resources to do 25 years ago.

[d. gauss]

 

Originally posted by MikeRivers Do you think that an experienced engineer is going to put his name on a "brittle" sounding mix or track unless he wanted it to sound that way?

 

 

unfortunately, it happens everyday. sheer economics force the trade off between sonics and ease of doing things quickly and in an easily repeatable manner. labels & producers often demand protools.

 

-d. gauss

[MikeRivers]

Originally posted by MitchG If I play you a CD of a song you've never heard before, you can tell if it was mixed in the box or through a console? Or you can tell in comparison to hearing other stuff? Have you tried double-blind tests?

Unless you were inteltionally trying to trick me, if you played me 10 CDs and asked if I thought they were mixed ITB (geez, I hate that abbreviation but it's what we're talking about so I'll forgive myself for using it) or on a console, if I said "ITB" every time, I'd probaby score better than 80% correct. Why? Well, because that's just the way records are made today.

 

It's like predicting the weather. If you say that tomorrow's weather will be exactly like today's, you'll be right 80% of the time. That's about as good as TV weatherpersons.

 

But could I identify one or the other positively? Not likely, unless it was due to an error that was unique to one process or the other. If the "sound stage is collapsed" you can turn the pan pots further and adjust reverbs and delays to compensate. If the low end is tubby or the highs are brittle, you can fix that with EQ. But you gotta recognize it and do something about it, not just leave it and say "well, digital mixing does that."

 

Perhaps Lee's position is that you can't - but I figure that if you can hear a problem, you can fix it if you have the right tools and the skills to use them properly.

[Dave Martin]

Originally posted by Lee Flier Interesting idea, but I think it's more to do with the fact that the rise of even tempered tuning more or less went right along with the rise of Western science and medicine. And once Westerners started "worshipping" science they began to doubt the power of music as a healing force, even though doctors back then pretty much didn't know jack.

 

Lee, you might wanna read this: www.uk-piano.org/edfoote/well_tempered_piano.html I had the author in yesterday to tune my piano; he's a most interesting person, and his take on well tempering is quite interesting (in fact, I think that I'll have him well temper the piano the next time I do a solo piano or jazz record).

[Dave Martin]

 

Originally posted by Lee Flier Usually I can tell without any other reference. Generally there will be something that jumps out at me as either "There is no way that's ITB; it has to be a console" or "That's gotta be ITB." Usually the giveaway is the reverb tails, or something at the extreme high or low end.

 

 

Here you go, Lee - http://cdbaby.com/group/cuppajoerecords

 

At least one of these records was mixed through the console - and at least one was mix in the box - but which method was used on each record? Admittedly, MP3's off of CD Baby don't give a lot of detail...

[RockViolin]

Originally posted by MikeRivers Fiddles have been around for 1000 years, and I still can't play one very good. But at least I have the good sense not to do it in public.

 

I sure would like to have a little talk with the person that first called it a fiddle.

 

The violin, in its current form has been around for about 536 years. Add another 100 or so for the viol. That's plenty of time to have picked up a lick or two. Sorry to OT...now back to yer own.... fiddlin.