(for PC's running Win 95/98/ME)reviewed by C.S. Graves
You must have seen those ads in Electronic Musician back in 1997, the ones that claimed a soon-to-be released software synthesis package could perform a great variety of synthesis techniques at a fraction of the cost of comparable hardware. And you may have read the glowing reviews in said magazine and others that seemed to support the hype. But how does Seer Systems' Reality really perform?
One night on the web, I decided to take a break from my usual downloading of pornography and downloaded the Reality demo instead. While I did not get to delve deeply into the architecture and possibilities with the self-expiring demo, I did get to see just how low latency it actually was. At the time I had an AMD K6 2 333MHz processor with 192MB of RAM, and I was quite pleased to hear almost no detectable lag from the time I pressed a key on my controller to the sound pouring out of my speakers. Native Instruments' Reaktor had intolerable latency on the very same machine, but to be fair, Reaktor has other strengths to its credit.
Back to Reality. I was impressed enough with the speed and what I could make out of its capabilities that I decided to shell out about $345 for it online (including exchange rates, shipping). Now that I could really get my hands wet.
Reality does indeed handle a number of methods of synthesis: Analogue modeling, four-operator FM, sampling, physical modeling, and modal. Apparently Soundfonts are supported as well, though I haven't tried this, never having used them to any great extent. The first three methods are actually lumped together in a fairly versatile architecture. You can select one of five FM synthesis topologies, apply analogue style waveforms and filters, and you can throw sampled sounds into the mix as well. Also, you can create zones which can contain different sounds in the current mode of synthesis. The basic waveforms are there, even a modulatable pulse wave, as well as several types of noise. A lot of parameters have assignable modulation sources, and there are 4 LFO's and 4 Envelopes per zone to choose from. The LFO shapes are sawtooth up, sawtooth down, triangle, pulse (also modulatable), and white noise. Envelopes can be looped in several ways if you choose, and their times can collectively be modulated by up to two sources. The modulation for pitch, or "coarse tuning" as it is called in the program, is measured in cents, which is handy for quickly making a sound glide from interval to interval from a range of 1 cent to 3 octaves. That maximum may seem low to some of us though, who have heard synths able to change pitch from infrasonic to near ultrasonic. You can also determine how many cents there are in each half step (notes on the controller/sequencer), allowing for sounds with microtonal scales and/or alternate tunings.
I remember hearing about physical modeling in the mid 90's and wondering what it sounded like. Reality helped answer that question, without me having to pay US$4000 for a Yamaha VL1. Physical modeling involves creating algorithms to emulate acoustic sounds. To my ears the physical modeling algorithms in Reality didn't sound much more realistic than many sampled sounds and considerably more CPU intensive (the Tibetan Bell program ate up nearly all of the 333's resources with just a few notes!). However, they are apt to take less disk space and be more modulatable in sound. And if you don't think they are true to existing acoustic instruments you can do as Hans Zimmer suggested, create your own "virtual acoustic" instruments.
There's an 8 partial additive mode called tonal synthesis, though you can't actually assign a different amplitude modulation source for each partial.
On top of all the individual modes Reality has, you can mix programs together in what's called a "Patchwork", where you can combine several programs and apply mix and pan levels to each.
Effects are also included: Reverb and Chorus. The chorus effect actually has presets that act as flangers and delay. One of them is really wacky sounding, though the rest are fairly useful.
The capture function lets you render sound generated by Reality as a wave file on your hard drive, which means one could make all-digital recordings, no noise, no 60Hz hum! Just make sure DMA is checked for your hard drive settings, or everything will come out choppy.
Now, since this is a software synth, CPU usage is going to be a major issue. Reality includes a CPU meter to see just hope many resources it is using up. Fortunately there are things you can use to help keep Reality from going beyond your computer's limits. You can set a maximum CPU percentage and polyphony. You can choose whether a sound will be polyphonic or monophonic. Sample rates of 44.1, 22, and 11 kHz are selectable for all sounds. Especially useful for drum kits are the "choke groups". You can assign certain sounds to various choke groups. Notes belonging to the same choke group will mute older notes as new ones are played. In spite of such steps, performance is ultimately dependent what shape your computer is in. On my 333, Reality driven by an external sequencer could get a few monophonic and percussion tracks with modulation, but that would eat up most of the CPU. I am glad to say that after upgrading to an AMD 1GHz Thunderbird processor and a clean reinstall of Windows 98SE, performance was vastly better. RoB Patey and I are currently working on a track sequenced in Cakewalk Express Gold. So far there are seven tracks rife with modulation all using polyphonic programs with sample rates of 44.1 kHz. And we're only using about a quarter of the maximum CPU usage!
This is not to say that Reality leaves nothing to be desired. Portamento for example. I know, it seems like too important an aspect of sound to be left out of an otherwise versatile program, but I couldn't find any way to get portamento out of it to save my life. And there are other, more obscure capabilities I would like to see, like sample start modulation, loop-position modulation, sample-and-hold LFO shapes, et cetera. Things that may not be as commonly used as portamento, but that would nonetheless add greatly to the possibilities. (sigh) Oh well!
As computing power becomes less expensive, a system running Seer Systems' Reality coupled with a sequencer and a MIDI controller is becoming a greater and greater threat to pricier hardware workstations that don't do quite as much.
(Though I still wouldn't mind seeing portamento implemented in future versions!)
