Let's look at some of the pros and cons of each plugin format. Many plugins that run on a PCI card obviously require the purchase of the card itself, yet another expense to stretch our already too-small budgets. The advantage of this card, however, is that it is purpose-built to run these plugins as efficiently as possible, bearing nearly all of the DSP computations and allowing the native CPU to focus on the host DAW's performance, while maximizing sound quality. Accordingly, a DSP card can often be used to process audio as you record it, meaning you could use a plugin in realtime to compress a live-mic signal. Trying such an operation with a natively hosted plugin could cause monitoring latency, which never helps the performer!

On the other hand, signal processors that use the native CPU are usually cheaper then their card-based equivalents and not only because you don't have to purchase the card. A good example of this would be Pro Tools HD's TDM format, whose plugins are regularly twice as expensive as the native versions. Additionally, native plugs are far more portable than card-based DSP, since many of them are authorized with an iLok or entry of a serial number.

Until recently, it was pretty much unanimously agreed upon by the audio community that most of the best-sounding plugins required a DSP card to handle the intense computations that made them sound superior. In the last couple of years however, with multicore processors becoming the norm in most new computers, a number of native plugins have emerged, boasting sound quality that some users claim rival the best in card-based DSP.

At the forefront of this development is Softube, a Swedish company of, as they put it, "rock 'n' roll scientists striving to make the most accurate simulations possible of professional audio hardware." Softube is becoming renown for their shockingly realistic software modeling of audio hardware, with their native plugin emulation of the famous TubeTech CL1B turning more than a few heads. "We take a very academic approach to analyzing and modeling the hardware," engineer Torsten Gatu explains. "A lot of development has been put into a system where the input - schematics and measurements - more or less yields an automatic result where every component of the hardware is taken into account. This is real modeling and a long way from how many others do modeling, by tweaking off-the-shelf algorithms." And perhaps this is how Softube is able to capture the character and not exactly linear nuances of hardware. As previously mentioned however, advances in multicore processors have also played a large role in the creation of such accurate plugin modeling. Gatu confirms: "Faster processors and more memory are a factor in the sense that native processing now is a very competitive option to DSP hardware. This means that companies that in the past would have done their products solely for DSP platforms, now do them for native instead."

Another such company would be Cambridge, Mass.-based iZotope. With a product line ranging from audio restoration tools to complex mastering processors, iZotope's approach to DSP seeks to improve upon its analog hardware counterparts, rather than replicate them in the box. "Some software companies do strict one-to-one emulations of classic hardware in software form," says project manager Nick Dika. "That's never really been our thing. In the case of [multifunctional, iZotope mixing plugin] Alloy, we looked at a handful of sought-after compressors for example, and ran tests on them, listened to them, then took the characteristics we liked and wrapped them into an algorithm with its own character and responsiveness. Rather than imitating one piece of hardware, we looked at the best parts of analog processing, combined them with the precise control of digital processing, and made something that sounds great - but is also far, far more flexible than any one hardware processor."

For iZotope, which also develops and licenses its technology to third-party clients, it seems that the focus is not necessarily on hardware vs. software (their ANR-B is a hardware noise reduction unit) or even card-based vs. native DSP - it's, as Dika puts it, finding "new and interesting things we can do with audio and packaging the new technologies into something that will benefit our customers." The technology itself is simply a means to an end. Dika concurs: "Increases in processor speed have always been a great way to open up more possibilities for our DSP. And faster processors and more memory don't just allow for more modeling of dynamic characteristics, sometimes they enable features that just weren't possible before like some of the approaches we take in our RX noise reduction and Radius time stretching. There's nothing quite like the sound and feel of using a classic console, true. But on the other side of the coin, there's nothing in the hardware world that can produce the wild effects that iZotope Spectron can make - it relies on technology that's not physically possible in the analog realm. I think that's what's most exciting about the future of DSP. We can take the best parts of the analog world along for the ride for sure, but it's the prospect of seeing and hearing things no one's heard before that is most exciting."