KDFX Algorithm Overview

By Blake Stone, JBuilder R&D, Inprise Corporation, used with permission

I've just spend the past two hours walking through every algorithm shipped with the KDFX, tweaking parameters and giving each one a good initial listen with a variety of instruments and settings. I am very impressed, but that's been said before on the list. Nobody has taken the time to dive into a more detailed subjective description, though, so I'm going to undertake just that task. Settle in for the ride, because there's lots to cover...

Focusing on Algorithms

Others have already covered the use KDFX to do multi-effects in a song (see KDFX Sample Song), and the technical nitty-gritty of what algorithm has what parameters are covered in the manual available from Kurzweil's FTP site. I've already posted a bit about the signal routing options in the KDFX, so I'm going to focus on the sound of individual effects. How do the algorithms stack up? How many PAUs do you need to get what you want?

Algorithm Types

There are 108 algorithms that ship with the stock KDFX unit, of which 106 are devoted to manipulating sound in one way or another. The two other algorithms provide visual feedback but don't produce sound. One is for stereo image analysis and diagnostic tool to give you a visual image of how real-time controllers are mapped to parameters.

The available algorithms are categorized as follows: Reverbs, Delays, Choruses/Flanges/Phasers, Combinations, Distortion, Rotary Speaker Emulation, Special Effects, and Studio/Mixdown Effects. I'll write about each category in turn, giving special attention to those of particular interest to me, and wimping out on some because there's just so much there! It's impossible to absorb everything in a single pass.

Before continuing, let me add that the KDFX manuals are quite explicit: youCAN load additional algorithms from disk. I don't know whether third parties are going to be given enough documentation to write their own, but Kurzweil seems to be planning to add some at a later date. It's something to look forward to, but I'm going to be very, very busy just trying to take advantage of what's there already.


There are 15 different reverb algorithms, with sizes ranging from 1 to 3 PAUs (you have 4 PAUs shared between the four stereo busses, and another 3 for the auxiliary bus... so 1PAU effects are very important if you want a wide variety available at once.)

The 1PAU "MiniVerb" is astonishingly good. I was expecting some audible compromises from the smallest reverb, but it's very warm, capable of extreme clarify and crispness, and doesn't exhibit any of the metallic ringing that creeps into altogether too many digital reverbs. It compares very nicely with my DP/4, and blows away the Digitech chip. The biggest drawbacks of the MiniVerb are limited control (you get individual early reflection delay times for L/R channels, decay time, room type, size, some basic diffusion controls and damping frequency settings, but that's about it) and a monotonous reverb tail for long decays. Ie: there's no "movement" in the decaying sound.

The bigger multi-PAU reverbs give much more control. Some offer gentle LFO based movement in the reverb tail, others give much better EQ control over the reflective space. Some are optimized for tight spaces, others for large halls. It's also worth noting that the MiniVerb is essentially a mono in, stereo out reverb. This isn't normally a big deal since the stereo image disperses very quickly in most environments, but few purists will like the _true_ stereo-in, stereo-out reverbs and since you always have room for a 3PAU effect on the auxiliary bus, this is a good place for such a beast.

Other goodies? The reverse reverb is great up to about 800ms and also keeps a delayed signal for an optional "hit" at any point in the reverb chain. Beyond 800ms the reverb buildup has some audible repetition for percussive sounds, even though settings up to 3000ms are allowed.

The dual MiniVerb is also nice touch. It's a 2PAU effect, and uses the left and right channels as inputs to two completely different MiniVerbs, each with its own settings. Why not use two individual MiniVerbs? Because this way you only tie up one of your four effects busses and get two reverbs... and since the MiniVerb input is actually mono anyway, all you lose is stereo placement of the dry signal.


Since the delays are 100% digital, you get pristine repetitions of your sound with up to 100% regeneration and high frequency damping control over the regeneration loop.

There are a variety of delay algorithms, including a 1PAU 4-tap stereo delay with a maximum delay time of 2540ms and a 2PAU 8-tap stereo delay with up to 5100ms of delay time. All of the delays offer the ability to specify delay times in beats, and sync to a specific BPM, or the system tempo (even slaved to an external sequencer!) This is a great touch, though I have yet to torture it with changing tempos to see how well it tracks them. I wouldn't expect miracles, but for steady tempos it saves all the math involved in getting your delay loop and taps synchronized to a given tempo.

The "spectral delay" is a nice addition. Available in a 2PAU 4-tap and 3PAU 6-tap configuration, each tap gets its own shaper, pitcher, level, and pan position. The shaper is similar to VAST's shaper, bringing out additional harmonics in a sound, which the pitcher can create a pitched sound of a specific frequency from non-pitched information like a hi-hat or snare.

Chorus / Phaser / Flanger

The chorus is pleasant and smooth, and offers pretty deep control. The delay until the chorused voice joins in is selectable from 4ms up well beyond the musically useful range, the shape, speed, and depth of the chorus LFO is adjustable, and a single 1PAU effect can offer two independent chorusing setups for the left and right channels of an effects bus.

Flanging and phasing range from subtle to overt, with 2PAU variations providing more simultaneous detuned voices (four with the 2PAU effects, only 2 with the 1PAU in the algorithms I dug into.)


When I first read about the KDFX's signal routing, I was surprised that you couldn't route one effects bus into another (with the exception of the aux bus.) After all, combining several effects in series is absolutely necessary for some purposes. Fear not, because the combination algorithms offer a pretty wide selection of 2- and 3-effect chains that can be set up in series or parallel on a single effects bus. Most of the chains that use only delay, chorus, and flanging in a fixed order use only 1PAU, while chains involving reverb are typically 2PAU effects.

I've found that many "multi-effects" processors have frustrating limitations on the order of effects. A delayed reverb _does_ sound different than a reverbed delay, and many boxes only give you one of the two. Again, Kurzweil comes through with many combination algorithms that let you pick which effect comes first. These are typically (always?) 2PAU effects, but the flexibility is a welcome addition.

Some of my favorites on a first listen are the Reverb<>Shaper (the "<>" symbol means that the order can be switched) the Quantize+Flange (which can produce lovely grungy effects) and the Reverb<>Compressor. The latter is one of the few 3PAU combination effects, but considering that it gives a full, lush reverb and a fully configurable compressor with sidechain EQ and 30 adjustable parameters... I'm not complaining. Compress the tail of that reverb for you? Let the rumble through, and only compress when the tail has high frequency content? Yes, sir!


I'm not the world largest distortion connoisseur, so I only spent enough time to find that all my needs would be served. There's a trivial 1PAU distortion algorithm that is quite limited, sounding more like an overdriven transistor than anything, but the 2PAU distortion was quite serviceable, and the 3PAU effects offer a range of tube amplifier simulations with cabinet emulation or tunable EQ, integrated delays, and tons of other goodies.The grunge could be warm or thin according to need, and a simple steel string guitar becomes a dangerous wailing weapon without too much work.

Rotary Speaker Emulation

Heaven. Absolute heaven. Kurzweil went all out on this one, and combined with KB3 it's quite an amazing beast. The simplest versions are 2PAU effects, but there's a 4PAU version and a full 7PAU version that uses one effects bus and the auxiliary bus together. You don't need to devote your entire KDFX unit to the task, though, as the smaller versions are pretty fully featured.

To be honest, I couldn't really feel much difference between the 4PAU and 7PAU speaker emulations, because the 4PAU version is so good to begin with. Doubtless there are subtleties that make it worthwhile if you're going to devote your K2500 to the task of _being_ a B3... but the differences are subtle. With the 4PAU version you can control microphone placement (two per speaker), crossover frequencies between the two speakers, speaker rotation speed and direction, chorus or vibrato, distortion, and cabinet emulation. Everything from sparkling clean sounds to deep and dirty sounded great, though they didn't go far as to emulate the sound of the leslie motors and the "click" of engaging them. I know... whine, whine, whine. :-)

To get away with a 2PAU version, you need to sacrifice either distortion _or_ microphone placement, vibrato/chorus, and cabinet emulation. If you can live without one or the other you still get an absolute world class leslie simulation with the rest of the bells and whistles intact.

Special Effects

I didn't spend much time with the ring modulation, pitcher, or super shaper. They all looked pretty nifty, and I'll doubtless find a need for them at some point. What I did play with for a while was the filter. Why? VAST has filters, too, so what's so special about the KDFX filters? They feel a little warmer, for one, but more importantly they're very responsive. The envelope follower and LFO "snap" from closed to closed with an expressive sweep that VAST can't emulate because VAST's realtime controls including envelopes are only measured 50 times a second. Here's where you'll get those analog-gear-of-yesteryear emulations down pat. And of course being able to apply processing to the sum of several voices adds a new flavor as well. Price of entry? 2PAUs for a single stereo filter, or dual mono filters.

Studio/Mixdown Effects

Similarly, I glossed over the tremolo, EQ (because there are already two bands of basic EQ per input channel in the KDFX and I didn't want to waste a single precious PAU on more EQ), skipped past the gate, and bypassed the stereo enhancers pausing only long enough to realize that the SRS encoder needs about a 5dB cut on output to avoid amplifying most signals. I spent a little more time on the compressors and the enhancer.

The compressors, in both soft and hard knee variants, are extremely responsive. They'll clamp down on a transient so fast it'll make your head spin, worked wonders in adding sustain to a variety of voices, and vary from extremely subtle at 1:1 (okay, so that's more than just subtle) to absolutely stonewalling at Inf:1.

You get pretty subtle control up to about 19:1, after which it leaps from 30:1 to 50:1 to 100:1 before going to Inf:1. Starting in an entirely digital domain with a lot of dynamic range to spare, I never heard a hint of grit even when boosting a highly compressed signal by 30-40dB. Control over the opening and closing time for the compressor is very granular, and you can even clamp down on fast transients with a gradual compression attack by adding a touch of delay to the audio signal, while the compressor uses the real-time signal to get advance warning of when to clamp down.

The basic compressor is 1PAU, if you want the controlling signal (the "sidechain" in compressor-speak) to be EQ'ed first you can get there with a 2PAU version. There's also a compressor that divides the audio into three different frequency ranges, compresses each separately and combines them for a health 4PAUs. I'm not sure what I'd need it for since it distorts the balance of the instruments I tried it with, but you never know.

I absolutely love the enhancer (or exciter, or whatever you want to call it.) It brightened up the sounds I tried it on with no effort, and none of the digital noise I normally associate with this kind of effect. The one I spent my time with is a basic 2-band enhancer and a 1PAU effect, but a 2PAU 3-band enhancer is also available.

Closing Notes

I'm very, very impressed. I love everything I've heard and can't wait to spend more time digging into the depths of the KDFX. One of the first things on my list is going to be real-time control. The few parameters I tried (compression threshold and tube drive) could both be swept in real-time without glitches, clicks, pops, etc. Very smooth, very usable, and very easy to set up.


- Blake Stone

Back to the K2500 Page

Back to the RavenWare homepage