Universal Audio’s UAD-1, a sound processing platform built on DSP hardware add-ons for your computer, has gotten a much-anticipated sequel this week. The UAD-1 was always a favorite choice for sound production, delivering tasty analog-emulating sound tools on a PCI card platform. The UAD-2, on PCI-express cards, offer up to “ten times” the processing power of the original — supposedly even the single-processor model delivers a greater-than-twofold performance gain. The DSP hardware is just the platform, though, and Universal’s main push here is its plug-in developers. Sure, these days your CPU is a plenty-powerful sonic number cruncher, so I think it’d be a stretch to say anyone needs DSP cards. But what the platform can mean is plug-in goodies not available anywhere else, with a no-nonsense approach to sound that may not be as practical in native plug-ins. (And with support from software like Ableton Live, Apple Logic, and Cakewalk SONAR, you can then drop these into your host of choice.)

The UAD-2 will mark the return of many existing plug-ins, like this Fairchild emulation. But you’ll be able to run more of them. And there’s new goodness on the way just for the UAD-2.

Here’s a look from around the Web at what people are saying about the UAD-2.

Oliver Chesler at Wire to the Ear notes what could be a real “killer app” / highlight of the UAD-2: a Moog multimode filter.

Here’s a pretty new plug-in for the new Universal Audio UAD-2! It seems to have all the right stuff too: self-oscillation, drive control, stereo tonal shifting, good modulation options and yay a wet/dry knob.

The Moog Multimode Filter for UAD-2 [wire to the ear]

Not to argue with the “classic design” or the genius of Bob Moog, but I do have to observe that the Fabfilter Product Line Oliver recommends, native plug-ins rather than Universal Audio, have more innovative interfaces that were actually designed for software. Don’t get me wrong — I might still have a great time with the Moog emulation — but this illustrates that CPU-based plug-ins remain competitive, and I’m not sure that emulating analog interfaces always makes sense on a computer. Then again, if you don’t have a rack mount Voyager lying around, I can’t argue with the appeal of a UAD-2 plug-in.

For more on why the sound aspect is so appealing, check out UA’s “realism” explanation (propaganda, yes, but worth a look).

TRASH_AUDIO have been eagerly watching this one for some time:

UA is promoting the fact that up to Four of the UAD-2 cards can run in one system, but just ONE Quad card will allow you to have 128 Neve 88RS channel strips open, which essentially gives you a 128 channel Neve console right in your DAW. I am upgrading my UAD-1 the second I find a place to buy the UAD-2.

Universal Audio: UAD-2, Out Now! [TRASH_AUDIO]

Key of Grey notes that UA’s digital hardware (UAD-1/UAD-2) reflects some really fine-quality analog gear:

Universal Audio makes some of the best hardware out there. I’m especially a fan of their 610 and 6176. The warmth of the analog sound makes a big difference when most of your stuff has that digital edge.

… The UAD-2 continues this tradition. Depending on how many tracks you want it to handle, you can pay for increasingly powerful add-on cards, even up to supporting 128 tracks of Neve console. Unfortunately, I don’t have a desktop to put these in but they present fantastic value for those who can’t afford tonnes of analog gear.

Univeral Audio UAD-2 : A much needed upgrade to the UAD-1 [Key of Grey]

But Can You Lift It?

Incidentally, those wondering about portability, a couple of options:

1. Get an SFF PC. I’m kind of curious to try putting a UAD into one of the two PCI slots available on my Shuttle, thus creating a “luggable” system with these sounds.

2. Get an Xpander/Xtenda. UA does make a product specifically for ExpressCard-equipped laptops like the MacBook Pro, so mobile is definitely an option (as it is with the rival TC|Electronics PowerCore). At the moment, I can only find the desktop/laptop bridge Xtenda product on the UA site. Updated: as confirmed in comments, it seems a laptop-compatible UAD-2 project is in the works as a successor to the UAD-1 Xpander product; we’ll keep you posted.

UAD-2 and Compatibility

I’m curious to find more about whether the UAD-2 introduces any new compatibility issues with either plug-ins or hosts. A number of host developers only recently got all the issues with the UAD-1 ironed out. My uneducated guess would be that these should “just work” with the UAD-2, but I honestly don’t know, so it’s on the top of my list to go research. Host developers, feel free to chime in, off the record if you must.

So, readers, who’s getting a UAD-2? Budgets are tight for a lot of us at the moment, but then, the UAD compares favorably with a lot of the pricier Pro Tools plug-ins, for instance. US$500 gets you a ticket to ride, with generous plug-in vouchers as you upgrade so you can build your own bundle. (If you’re feeling poor, stay tuned for some Recession Special coverage coming your way soon … but UAD lovers, I’m sure, will sell their car before they miss a chance for a new UAD.)

Universal Audio Site with all the specs and whatnot

  • Darren Landrum

    Not too long ago, I wanted to put together an open hardware project, like the Arduino, to make our own DSP PCI card that anyone could then code for. It was then pointed out to me the hundreds of thousands of dollars in tools, not to mention the hundreds of hours of expertise, required to make a PCI card. I'd thought about going Firewire instead, but never really looked into it.

  • Aragon

    I would wait to get one of these, they're going to get competition from Nvidia CUDA, graphics card are becoming a incredible fast general purpose computer card. The Nebula3 vsti is the first, I guess, optimized for it.

  • Adrian Anders

    I think much of the appeal of these cards/external DSPs are legacy hardware developers who would otherwise not enter the market unless there was a piece of kit that made piracy impossible (and thus allow them to price their plugs at an extraordinarily high price). If a cracker is determined enough, you can crack a dongle. But an external DSP? Call me when you've cloned the Powercore, Scope, UAD, Plugiator, and Vari-OS on the PC.


  • @Aragon: GPGPU (processing audio on the GPU) is really a subject for another article. But I don't really see it as competition for CPUs, let alone dedicated DSP, any time soon. GPGPU programming is hard. DSP programming is also hard, but part of the appeal of DSP platforms from Pro Tools to UAD is that many audio manufacturers are already used to doing this kind of DSP programming. (Some of it can be ported fairly easy, as well.)

    GPGPU starts to get interesting for massively parallel processing tasks — possibly granular synthesis, for instance, or physically-modeled synthesis that was previously impractical. But I'm extremely dubious about having to target CUDA as an emerging platform that doesn't work on all machines. I think this stuff could really take off with OpenCL, assuming it works and is roundly embraced by hardware manufacturers. (Apple is backing OpenCL, but then the issue is really the hardware makers, not the OS/computer vendors.) The outlook there looks reasonably good — but we'll have to wait. And I doubt it'll look anything like the UAD.

    Even CPUs have a tough time competing with the cost equation of the UAD, given you can add a card for $500. So I think there's going to be a place for CPU/DSP hybrid solutions for some time to come.

  • Any word on a price drop on the UAD-1? If I could afford to put 4 UAD-1s in my G5, I'd be a happy man.

  • beej

    Yes, the FAQ on the UA site states that there is a laptop UAD-2 solution in the works – ie a UAD-2 Xpander – apparently it's due this year…


  • @Darren Landrum, making a firewire device is going to be even harder and more expensive than making a PCI device. I would also venture to say that making a USB2.0 HiSpeed device would also be harder than PCI.

    However, whoever told you that you needed hundreds of thousands of dollars in tools was dead wrong. You can design a PCI card using the free(speech) PCB program from the gEDA suite. You can program the xilinx FPGA using the free (beer) Webpack from Xilinx's web site. And you can even find a guide to help you get started with PCI on fpga4fun.com.

    As for a DSP, I'm sure there are reasonable choices that don't cost too much and aren't too complicated, but I'm not familiar with them off the top of my head.

    You are write about the expertise to some extent. Starting on a PCI card with no design experience probably is a mistake. However, it is fairly approachable with a little experience designing your own PCBs for other purposes.

    If you are interested, try designing and making a few

    shields for the arduino, then try making your own arduino clone based on one of the larger AVRs, preferably using as many surface mount components as you can.

  • gbsr

    @Darren: i would be very interested in this (interested as in getting one myself).

    @peter, free beer webpack? im drinking beer right now, its not webbased and it definetly wasnt free.

    @the article: yummy. if only i had the money for it. but like you said, since our computers are highspeed monsters these days theres really not a compelling good reason to get an external dsp card, but id probably get one just in case given the money then.

  • Darren Landrum

    Well, to be honest, I really do think things like CUDA and OpenCL are going to be better options for hardware-accelerated synths and effects for the open-source world. In fact, it was when I first broached the idea of an open DSP platform that someone pointed CUDA out to me.

    As for making the card, well, the guy who told me these things works as a digital designer for TI, so I would assume he knows what he's talking about. Still, someone somewhere was able to design the Arduino, so things like this must be possible, right?

  • @Darren — CUDA, OpenCL, we'll just have to see for the long run; I'm talking about the short term here. And for that matter, if they work as development platforms, it might appeal not only to open source devs but commercial devs, as well. I still think, though, that a certain amount of inertia applies and perhaps should. DSP chips and CPUs do their job now pretty well; OpenCL gets most interesting when you get into those areas (and there are quite a few) in which DSP or general purpose CPUs *aren't* optimal.

  • I don't know if CUDA is going to improve on nvidia's first round of "GPU as DSP" design, but that first round had the fundamental problem of massive processing power coupled to major latency. Very cool if you want to do ray tracing or even live video where 1/30 second is available for roundtrips. It doesn't work for audio, most of the time. CUDA might be different, but I just wanted to mention this issue.

    On the UAD-2 itself, I find it comical the level of enthusiam that is thrown at these add in DSP boards. Back in 1989, I was working at a company in Seattle where we got several DSP "daughter boards" for doing PostScript rendering. They clocked in at about 20MFLOPS which was considered massive at the time. Ever since then, I've watched announcement after announcement about how a new DSP generation is going to finally [fill-in-blank-here], and yet they seem to never really take off in any significant way. Creamware? Chameleon? Even the UAD-1 and PowerCore, which appear to have been successful products by some standards, remain tiny niches in the already tiny niche world of pro-audio plugins.

    Its great that the smart people at UAD have a large enough market to justify the R&D that they put into these new devices. But lets not kid ourselves about any particular instance of this type of hardware having anymore significance than the release of the first 1GHz x86 processor. Hell, I recall the first 10MHz, the first 100MHz, the first … well, you get the picture. If this tool helps a few people do something they need to do, thats great. But DSP-tied platforms are not the future, and everybody knows it.

  • A friend of mine asked my opinion of the UAD-1 back when he was setting up his first home studio 2 years ago. I told him to be careful because once you switch computers in 3-4 years, the cards might not be compatible with his next computer. Turns out he bought a laptop recently.

    I also kind of agree with Paul Davis (thank you for your free software btw). It's not really mindblowing and in the future the interest for those platforms might be low as cpu power keeps on increasing.

    However, i can see many cases which it makes sense to buy UAD cards. A mid-range studio which needs lots of power to run plug-ins might see the UAD as a good solution. Protools is so incredibly expensive from what i can see that one could load up a mac pro with uad quad cards and come out the same price as a basic protools setup (talking about the HD version, not the simpler home studio versions of protools)

    If you need lots of power today, these cards seem like a good solution. I even am tempted by those (but have no money). I have no experience with recent workstations like the mac pro but i expect that you could already run a fair amount of regular plugins on a machine with 8 cpu cores. Some people might want even more than that.

    You can also expand your system one card at a time, which is quicker (and perhaps less expensive than upgrading a computer.)

    One thing i wonder about the Open CL and CUDA thingies… What happens if you need the gfx card ressources in your audio software ? Perhaps some day 3d graphics will be common in audio software. Perhaps for some audio analyzis or for different user interfaces. Once you load up the gfx card to run 2 displays and a bit of 3d, won't it reduce significantly what you can do with these new technologies ?

  • kj

    @paul davis

    if you don't think the UAD-1 card was a 'success' you *clearly* know nothing of pro audio or you're just hatin' or both.

  • Well, Paul, I don't disagree, but it sounds like you answer your own question. This is hardly a quantum leap in processing power, but for that niche market, it does some nice things and apparently provides value (at least to that niche).

    As for GPGPU, this stuff is very, very new — it's hard enough to get solid answers on some of these questions, which is why I suspect applications are further off. But a few quick points:

    First, real-time DSP is only one audio application; I could see benefits to non-real-time applications, including for audio. (Personally, I'm especially interested to see what happens with computer vision, as that's a *lot* more efficient and doesn't encounter the round-trip issues.)

    I'm honestly not certain on the round-trip latency with CUDA 2.0 or what's happening with OpenCL; I'd say at the very least, stay tuned, as the architectures are shifting. AMD/ATI are building hybrid CPU-GPUs, so that the traditional bottlenecks between the CPU and GPU are disappearing. Intel is trying to build better graphics chips, again closer to the CPU. NVIDIA is doing some of the same.

    But secondly, audio GPGPU applications robbing from display resources is NOT a problem. The whole appeal of GPGPU in consumer computers is that a lot of the graphics power of your machine goes untapped most of the time, unless you're playing games, etc. Even with the GPU accelerating display routines on Vista or OS X or 3D windowing environments on Linux, there's plenty of leftover horsepower, even with older graphics cards, let alone the ones we're seeing now.

    That doesn't change the bottom line, though — you use what's available now. True, DSP chips haven't exploded in performance — but then, they haven't become outdated as some might have predicted, too, particularly for these audio tasks. So the UAD cards still remain a decent value for now. My own preference is generally for native plug-ins because of their flexibility, but I won't argue with the equation some people see with the UAD. How that changes in the future is another matter.

  • It may be because this is a particularly techy blog, but I think many are missing the point… I agree that addt'l DSP becomes less and less relevant, but the UAD plugs are really great, and even with the cost of the cards factored in, they're still a FAR better value than competing Waves products. You really do need to think of the cards as a dongle that actually has some benefit (instead of actual dongles that frequently just screw things up).

    Obviously this won't be everyone's preference, but Logic with a serious audio system (Apogee, Lynx, RME, etc.) and one of the new UAD cards makes for a formidable audio system at about 1/2 the price of a PT TDM setup (provided you aren't attached to PT software).


  • @Mitchell: Well, Paul Davis as the author of JACK and Ardour and one of the foremost voices for Linux audio gets the right to be a little biased toward the solution he's put his lifeforce into. And obviously, the model there is pretty far from the one here.

    We just got a little off-topic on GPGPU stuff. But it's a question worth asking. Originally, platforms like the UAD were *necessary* — that is, it wasn't possible to do the processing on the CPU. Then, it became simply more luxurious to off-load some tasks from the CPU or to have to cut fewer corners with implementation. Now, I think the situation is different, which is that you get specific sound tools you wouldn't get otherwise.

    By the same token, the GPU-based processing in regards to audio *is* well into the future. Still, so were these kinds of DSP chips and CPUs at one point. So it'll be interesting to see what happens.

    You're absolutely right in that the UAD is a *not-in-the-future*, excellent solution for some right now.

  • @kj – i am entirely aware of the status of the UAD-1, at least to the extent that anyone outside of UA can be. my definition of success is, i suppose, a bit more expansive than yours. the UAD-1 has been a very successful product within its niche market – my point is that its niche market is a subset of an already tiny niche market.

    @mitchell – great perspective. its true that for software creators a hardware "dongle" is a very useful way to "protect" your product and i totally agree that in this sense the UAD approach is more useful to the user than the waves one. the pricing of both, however, limits the market for the product, and the UAD approach comes with a certain amount of built-in obsolescence. within a couple of years, maybe more, your UAD-n card will do less than your CPU(s) can do. the waves model, where the dongle is just a pretty simple USB challenge/response key, continues to benefit from all the R&D being put in by Intel, AMD and others, and your dongle never becomes obsolete (theoretically speaking, anyway).

  • Without getting into GPU and other discussions, I'd like to quickly share that I used the UAD-1 plugins and I was seriously impressed.

    Alright, it's 5am, I should go BACK to sleep 🙂

  • I'm not sold on GPGPU stuff myself. (And I just came back from NVision08 last week, where there was a lot of CUDA talk and a bit about DX11 "compute shaders." There was a distinct lack of audio-related discussion.)

    I get the sense that it's kind of a hacky stopgap, like Commodore 64 programs that took over the processor on the floppy drive. I think we're going to see a bit less of an architectural schism in a few years; maybe the GPU will go the way of the separate FPU, and CPUs will have this massively parallel vector processing monster just built right in, and you can drop a new one in every 6 months for $300 if you want to.

  • poopoo

    It's such a pity the plugin's look like tired old emulations of fetished gear. The so called "killer app" is a low pass filter?!?

    I'd rather see something like the Korg Oasys PCI only updated for the 21st century.

  • @poopoo: Well, I called it a killer app, just because people liked it. But I don't think it was "I need that. I'll buy a UAD-2." I think it was more, "If I owned this, I could run the Moog filter" (on top of everything else).

    But yeah, I think a lot of UI design has moved on from simply copying vintage gear. Fortunately, there are some people doing amazing work in that area in software.

    OASYS… interesting. Well, the OASYS as you know runs natively on a CPU (hello, Linux). Which brings us back to the original question here, eh?

  • YJ

    i had an UAD-1 PCI-e card, but the UAD-2 costs

    $2000 per card, which is more than my three audio studio computers..

    bhahahaha what a joke.. i guess we have to wait a few years for the prices to come down..

  • Pingback: Create Digital Motion » CPU vs. GPU Mythbusters Demo Reveals a Lot()

  • Hi Peter, thanks for linking KeyOfGrey, I'm a big fan of this site. I'm certainly not an expert on add-on DSP cards, but I do most of my work on a MacbookPro and would welcome the ability to have processing power on the go. This is not to say that I would rather have this card over a cheaper and faster solution, like a Logic node, when I'm mixing which is normally at home. Nor, would I rather have this card over the emulated analog gear. At 2 grand for the fully optioned UAD-2, it's encroaching on the territory of a real Neve channel strip. I would gladly live with the "limitations" of only one strip if it meant I had the real thing. Still, the UAD-1 had some nice software plug-ins and there's certainly value if a person pays that two grand once for the hardware, and then is able to leverage the value of the plug-ins. In my opinion, it is what software developers make of the hardware standard offered by the UAD-2 that is most interesting.


  • Pingback: links for 2008-09-03 « Jay’s blog()

  • poopoo

    Barry Vercoe, the guy who wrote CSound, was a consultant with Analog Devices in the late 90's. They had a pci dsp card with sharcs that ran "extended csound" on the dsp. I wonder if any of the csound stuff is still used in the UAD?

  • Perry Sterry

    Pro Tools users, don't waste your money. Serious integration issues. UAD cards will not run other plugs. If you only have slots for three cards, you'll have to sacrifice one of your PT cards and intern HD processing power. An expansion chassis will cost you about 2K. All this for what? Take your money and invest in some great plugs already available for TDM and in most cases, better than UAD plugs. Also, if you ever need customer service-You can forget Universal Audio.

  • Hello.

    I the oldest day here.

    To all intents, it is intriguing,shame thriller..HERE http://digg.com/news/entertainment/demi_moore_nud…. What tell?