Ardour + Harrison + OS X = Mixbus

fedexnman · Post by **fedexnman** » Sat Feb 19, 2011 7:01 am

JWL what # version (9.10, 10.04, or 10.10) of ubuntustudio are you using ??

JWL · Post by **JWL** » Sat Feb 19, 2011 9:58 am

The newest one, which at the moment is 10.10. 11.04 will be out in a couple months....

JWL · Post by **JWL** » Sun Feb 20, 2011 8:34 pm

Starting to dig a bit into MixBus. Almost ready to testdrive it on a production.

One thing I've noticed. There's been a lot of debate about Harrison's summing technology. As Ben Loftis writes, it isn't an emulation of Harrison's summing, it is Harrison's summing.

Some aren't convinced that this matters much. Count me among this group.... while I've thought I'd heard improvements over the years, I doubt I'd be able to pick many out in double-blind tests.

But MixBus has SERIOUS HEADROOM. Seems almost like it's impossible to overload. I'm sure this isn't true, but when I crank sound out to the buses, and crank the tape sat settings, it sounds.... analog. No digital ugly distortion at all.

I'm kinda speechless at how good this is...

Galen Ulrich Elfert · Post by **Galen Ulrich Elfert** » Mon Feb 21, 2011 11:14 am

It turns out that computers and software actually multiply, add, subtract big huge numbers differently. Remember when the p3 first came out and some mathematician noticed that it was dividing shit wrong?

A skilled programmer can work around cpu and programming language limitations to get a "good enough" answer. But that can mean different things to different people.

I'm sorry but that's a crock. 2+2=4 on every computer ever divised, and no matter how big the numbers get (at least within the realm that's applicable to DAWs.) That's what makes it a computer. It's the single most important thing a computer does. The pentium FDIV bug was a flaw that resulted in a recall for precisely this reason, even though it only affected 1 in 9 billion operations.

The summing "engine" in every digital mixing system is an addition operation. That's what "sum" means. And every gain stage in a digital mixer is a multiplication. You can add a bunch of other digital processing to simulate the distortion and non-linearities of analog gear, but those are totally separate processes.

I'm sure it sounds great, and there's probably a lot of really complex processing going on in there, but to say that MixBus IS Harrison's mixbus is fatuous. One is an analog console, the other is a piece of software. So is the use of terms like "headroom" and "bandwidth" when talking about software. The closest equivalents would be bit depth and sample rate, and their effect is different

Seablade · Post by **Seablade** » Mon Feb 21, 2011 12:17 pm

Galen Ulrich Elfert wrote: I'm sorry but that's a crock. 2+2=4 on every computer ever divised, and no matter how big the numbers get (at least within the realm that's applicable to DAWs.) That's what makes it a computer. It's the single most important thing a computer does. The pentium FDIV bug was a flaw that resulted in a recall for precisely this reason, even though it only affected 1 in 9 billion operations.

Actually you are wrong.

To show a very basic example that I did elsewhere...
http://www.gearslutz.com/board/5222115-post1669.html

Float based math on modern computers has an inherit inaccuracy with it, but at very tiny levels. Most programmers work around this successfully however when it would make a difference, what I posted above is merely a very basic example of what is needed to work around.

The summing "engine" in every digital mixing system is an addition operation. That's what "sum" means.

Not quite. The summing engine in any digital mixing system generally refers to at least all applicable math that happens on a signal as it is mixed to various places. This could be addition(For simple adding of two signals to each other) but often also involves many other factors, multiplication/division/etc. For instance to convert from dB to linear values, to get the modifier to apply to a particular signal.

The Mix Engine as Harrison stated refers to this as well as all processing done on the signal in the form of DSP to modify the signal characteristics. This includes EQ, Tape Sat, etc. and in particular the gain staging between each element.

I'm sure it sounds great, and there's probably a lot of really complex processing going on in there, but to say that MixBus IS Harrison's mixbus is fatuous.

Actually it really isn't. The summing engine like many things in Mixbus is part of Harrison's DSP, part of why you can't route it via Jack like you can in Ardour. However you are correct in that summing 1+1 will equal 2 in Mixbus, much like any other DAW if you are not doing any processing, panning, or mixing whatsoever on the signal.

One is an analog console, the other is a piece of software. So is the use of terms like "headroom" and "bandwidth" when talking about software. The closest equivalents would be bit depth and sample rate, and their effect is different

Headroom and bandwidth is perfectly acceptable to me actually. There is only a limited range of values that can be stored in a given memory space. Therefore you have limited bandwidth and ensuring that you have plenty of room to ensure no loss of data occurs(From clipping or other means) is ensuring you have proper headroom as well. Both of these translate well from analog consoles to me.

Thanks for the response. I'm really interested in this. Is Harrison implying that the sound quality of the MixBus ITB summing is superior to that most of us are already familiar with? Have they made a "breakthrough" of some sort?

I'm not trying to be a dick. They're just vague about it in their marketing materials. They call it "True Analog Mixing"? so who knows what the fuck that is.

What Harrison was really trying to get out was more the integration of every component of the mix into the UI in a way to more truly represents an analog console. That isn't to say there haven't been flaws over the years in some programs, but rather that they feel that mixing in Mixbus will provide a smoother and faster experience to get the finished product without any of the aforementioned flaws. So the end result, which I have found very consistent, is that you will get to a nicely finished product much faster. It is amazing how slow it can really feel to open a window every time you want to adjust the EQ, or compressor, or tweak down the makeup gain, add a little amplification to the track, etc. Not that you can't still do that, you certainly can, but the goal of Mixbus is that hopefully you won't need to nearly as much, and I have found that pretty consistently true in my experience, as have many others.

Before it is asked, from the FAQ on Mixbus...

What are the "well known flaws" of other workstations?
It is our opinion that the "gross" defects in many workstations include internal clipping, lacking dither stages in the DSP processing, multiple format conversions, out-of-control gain stages causing plugins to work outside their intended range, routing choices that cause latency/timing errors, Inability to see meters such as compressor gain reduction without opening the plugin dialog, bad ramping of plugin coefficients, and poor user-interface integration.

Of course different workstations will exhibit these problems to different degrees. Our goal was to design a mixer using the "best practices" that we have developed over the course of 30 years. Multiple subtle design decisions, accumulated over a long history, are required to make a truly world-class mixing engine.

And since I haven't posted here before, the standard disclaimer applies: I am in no way connected to Harrison Consoles. I beta-test Mixbus for them, but that is about it, and am not on their payroll in any way shape or form so take my posts with a grain of salt should you choose:) I am involved with the Ardour project on which Mixbus built though not as a regular contributor.

Seablade

Snarl 12/8 · Post by **Snarl 12/8** » Mon Feb 21, 2011 2:15 pm

Galen Ulrich Elfert wrote:I'm sorry but that's a crock.

It is not. You, sir, have crocked. Have you ever programmed anything having to do with even fairly large numbers? Did you double check your programs output to make sure it wasn't a "crock?" Adding two 24 bit numbers could give you a non-24 bit number, then what do you do? And you didn't address dithering, which is, actually, what you do do. I submit to you, that a 24 bit (32bit internal processing anyone?) sample that has been "filled up" is not a small number at all. Adding 24-128 (tracks) of them, in real time (96 thousand times a second) doesn't allow you to go crazy with the error checking and converting variables from one type to another all the time. You've got to have an efficient and very well designed algorithm to make some tradeoffs for you. Especially if its got to hand back an "answer" at the bit depth of all the given inputs. Even Javascript (especially Javascript) fucks up simple math once the numbers get big enough. You have to be careful to use the proper data types and not exceed their capabilities and/or be comfortable with how they round, etc. Chips are just software that can't be changed. So yeah, they're fucking up to, somewhere in there. Have you designed a chip that can do any calculation flawlessly? Any?

Galen Ulrich Elfert wrote:2+2=4 on every computer ever divised

This is probably not even true. I bet a computer scientist could easily divise[sic] a computer that flubbed this calculation.

We seem to worship computers as examples of perfection in our society. But they are tools that are only as perfect as their divisers [sic] who are human, and therefore fallible.

Galen Ulrich Elfert · Post by **Galen Ulrich Elfert** » Mon Feb 21, 2011 3:08 pm

Once again I'm humbled by the level of discourse on this board. Thanks, Seablade, for that fantastically informative response.
Of course there are errors in digital signal processing. I was way off base there. I guess my problem is with the way a lot of marketing tends to confuse the properties of digital and analogue systems. Best practices on an analogue system often translate to worst practices on a digital system (for instance, running levels as hot as possible before clipping).
My understanding is that the noise introduced by dithering on a high resolution (24bit) system is orders of magnitude below the noise present in even the highest quality analogue systems, and that should translate into a very different set of practices. I feel like many of the failures of existing DAWs have as much to do with bad user practices as with design (not that there aren't probably major design problems as well, and in a sense, shame on the manufacturer twice, for not better informing users/producing more intuitive interfaces.) The question is, do you teach people how to use digital systems properly, or just invent a digital system that behaves like analogue? I'm kind of on the fence.

chris harris · Post by **chris harris** » Mon Feb 21, 2011 7:07 pm

Have you tried Mixbus?

roygbiv · Post by **roygbiv** » Mon Feb 21, 2011 10:45 pm

yes, this is a basic problem with computer math, the dreaded "round off error". Been around since the beginning of calculators.

now, regarding this quote:

Galen Ulrich Elfert wrote:... The question is, do you teach people how to use digital systems properly, or just invent a digital system that behaves like analogue? I'm kind of on the fence.

I'm not. I think we need to design things that work the way people/musicians want/expect them to work, not computers/engineers.

I came to this conclusion at the beginning of the 80's, when I worked in a store selling synthesizers. Unfortunately, all the cool American analogue synth makers started dying off to the latest cheaper, "clearer" keybaords coming from Japan.

Not that the Japanese synths weren't useful, they sounded clearer, and they were cheaper.

But gawd it sucked to try to squeeze a groovy new sound out of them after a bud and a Bud. (DX7 - blech!)

Same thing happened with effect pedals - honestly, would you rather fool around getting an inspired, spacey sound out of a Roland Space Echo or some mid 80's Lexicon ?

Music is made by analogue creatures, ergo I think it is reasonable equipment makers should try and adapt to analogue properties and thinking, even if imperfect, should be retained whenever possible.

Even if some of that thinking arose merely to adapt to analogue limitations (QWERTY keyboard, anyone?), why try to force people to change years of thinking and intuition because there is a "better new way".

Sometimes the old ways work just fine.

Snarl 12/8 · Post by **Snarl 12/8** » Tue Feb 22, 2011 1:00 am

But at some point we're just holding on to analogies (pun intended) because it makes us old fuddy duddies feel secure, but we're robbing the kids of ever knowing what computers can really do.

Galen Ulrich Elfert · Post by **Galen Ulrich Elfert** » Tue Feb 22, 2011 12:14 pm

Yeah, we'll have to agree to disagree there, roygbiv. I'm much more interested in what Radiohead, or James Blake are doing with computers than in someone making widdly noises on an overpriced minimoog. You'll note both those examples make brilliant use of the oldest instrument of all, the voice. But they place it in a totally exciting and engaging context that is distinctly of their time and place. Likewise, I'll take an early Deerhoof album, recorded with Pods and Pro Tools Free, over a tired blues riff played on a vintage AC30 into a tape machine almost every time. Not because of the tools they used but because it's much more interesting and exciting music.
Sorry to take this thread off track. Mixbus does look really cool, and I hope to have time to try it out soon.

roygbiv · Post by **roygbiv** » Tue Feb 22, 2011 6:15 pm

no disagreements Snarl or Galen. My "get of my lawn" rant may have confused the point of my previous post.

I love what digital (computers) + analog (humans) produce - for example, have you ever checked out Beardyman with a Kaoss Pad? now that's some crazy shit!

http://www.dailymotion.com/video/x9xwui ... -kp3_music

My point was to question about whether digital should emulate analogue.

Since analogue methods 1. don't t emulate digital well, and 2. are more established, work flow wise, my rant was in response to your comment:

"... The question is, do you teach people how to use digital systems properly, or just invent a digital system that behaves like analogue? I'm kind of on the fence."

My point - I firmly believe digital should behave in an analogue fashion, i.e., when you want to change a sound (filter, attack, volume, whatever) you grab a knob and change it. The design of the Kaoss pad does not require our bearded friend in the link above to "think digital", it is letting him, as a beat-box artist, combine his voice with the use of his digits in a very analogue way to make some crazy music.

The digital thinking I hate is the (early digital) Japanese synth paradigm: "to change sound parameter, please kindly push button two, increment down to level three, then while firmly pressing button 2 and 3 simultaneously, page over to sound bank six by alternating pushings of button 3. This simple task will readily make it possible for you to increase parameter 4 by your intended value. Please enjoy in this way the changings sound experience"

When you mentioned "teach people how to use digital systems properly", I got bad flashbacks to those early Yamaha digital synth days.

Frankly, we the music makers and recorders should NOT have to remember things like dBV to dBFs conversions, etc. everytime we want to make music. Some of that should be pre-engineered into the equipment/software.

Just like the QWERTY keybaord is kind of silly, and originally designed to overcome analogue limitations, I don't want to retrain on the DVORAK or whatever "better" keyboard. There is just too much muscle memory, and typing instructions, etc. to turn away from QWERTY.

That, I guess was my point - there is so much old school training and thought that went into analogue systems (concepts like headroom, etc.), it makes more sense to me to simply make digital behave like analogue.

[Sorry for the thread Jack Harrison guys.]

Snarl 12/8 · Post by **Snarl 12/8** » Tue Feb 22, 2011 8:16 pm

Hey Roy,

Why does it have to be either/or? I like living in a complex ecosystem where the big lumbering dinosaurs have the fast nimble mammals running around as well.

Even though I'm of the lumbering dinosaur variety, it's an interesting time to be choosing a recording axe.

c

Galen Ulrich Elfert · Post by **Galen Ulrich Elfert** » Wed Feb 23, 2011 11:57 am

Ok, I see what you're saying, and I basically agree. The only thing I would take issue with is that I don't think analogue devices are inherently more useable, so much as they benefit from having been around for a lot longer. I know what you're saying about those early digital synths (I have a TX81Z), but I'm pretty sure The RCA Mark II that Milton Babbitt used at Princeton in the 50s was at least as much of a pain in the ass to program. And Kaoss Pads and looper pedals are examples of totally digital devices that are extremely intuitive.

roygbiv · Post by **roygbiv** » Wed Feb 23, 2011 2:43 pm

Galen, thank you, that is exactly the point I was (am?) trying to make.

Analogue systems are at the moment, in general, more usable mainly because they have gone through a greater amount of "tweaking" over the years to increase their usability.

Hence, their apparent user-friendliness now - they have been subjected to evolution over time to keep you from doing stupid things, and to make things more easy to do/intuitive.

I worry that all of that energy/thought/creativity is sometimes ignored with the creation of digital devices. Often times back in the day (and even now) digital-based systems appear to throw out all of that analogue engineering, claiming it is because "they are digital, you should learn digital", when in fact, the are simply trying to save a few bucks in manufacturing costs (I mean, how much more can it cost to put a few rotary encoder knobs on something?)

Anyway, I totally agree that digital can be made very user friendly. Realistically, it could be MORE user friendly than analogue ever can be, IF the designers want to include that in their design.

That is what triggered my initial not-well forumulated rant - I was responding to the question of whether folks should try to make digital behave more "analogue" like.

It's also what drew me to the original post - I'm curious about the Harrison Mix Bus - although I'm a Reaper fanboi and have used it for years (in part because of its analogue style routing), I am really curious interested in the (analogue style) total channel strip in the Harrison console program.

Can anyone chime in who has experience both with Reaper and the Harrison mix bus - pro's, cons?

Message Board

Ardour + Harrison + OS X = Mixbus

Who is online