Reality PC

Take a powerful PC. Add a clever bit of programming, a soundcard and some samples, and you have the Reality software synthesizer. Not a revolutionary concept, but wait until you hear the sounds! Martin Walker picks himself up off the floor.

This took me by surprise. Once I'd installed the Reality software, and plugged in a MIDI keyboard, I lost a number of hours in the twilight zone. There are some gorgeous sounds in here, and it only takes a few seconds to forget that you're actually playing a PC soundcard. I wasn't expecting love at first sound, but that's what happened.

The name Sondius may seem familiar, and if I mention the AWE64 Gold soundcard and WaveSynth/waveguide synthesis, you'll probably remember the connection. Sondius provided the software technology for Creative Labs to add some basic physical modelling sounds to their latest soundcard, for more realism and expressive playing. I'm sure everyone who bought a Gold card tried these out, and they were certainly significantly better than the equivalent wavetable sounds, but not exactly jaw‑dropping, and a bit of a fiddle to use. Well, if you think of those sounds as an appetiser for the real thing, Reality is most definitely the main course.

Reality is a software‑based synthesizer (fully 16‑part multitimbral and 64‑note polyphonic), incorporating Sondius technology, and developed by Seer Systems. This Californian company has Dave Smith as its president (the person behind MIDI), as well as an impressive list of credits. Unlike the first waveguide synthesizer, Reality uses an open‑ended system which currently allows waveguide (physical modelling), subtractive (analogue), FM, modal (using a bank of resonating filters), and PCM (sample) techniques, to create a much greater variety of sounds. The open‑ended design allows new features and further synthesis methods to be easily added — a vocoder has already been mentioned.

No doubt, at this stage, many people are expecting the usual scenario — create sounds on the computer screen, press the Synthesize button, and then wait several seconds at least while the sound is 'rendered' down to a WAV file, so you can play it via MIDI. Well, you're wrong. Reality is far more powerful than that — everything is created not only in real time, but with low latency (the time delay between pressing a key and hearing the sound). As soon as I started playing Reality sounds, it was just like having another synth module in the rack. There are even reverb and chorus built in!

What Do You Need?

Seer have worked closely with Creative Labs to enable such low‑latency performance, and for this reason Reality only currently works with Creative 16‑bit soundcards (SoundBlaster 16, 32, AWE32, AWE64 and AWE64 Gold). This is because the Reality drivers hook into the Creative audio drivers at a fairly low level. The company are working on many other soundcard options at the moment, and some additional drivers are expected by November. Of the Creative cards, the AWE64 Gold is the recommended option, since this has S/PDIF digital output, which allows higher‑quality performance to be achieved by using external D/A converters.

Since the synthesis is performed in real time, a relatively powerful processor is needed — at least a Pentium 133MHz, and preferably faster, and a minimum of 24Mb of RAM. For once, having MMX actually makes a significant difference. The higher the speed of your processor, the more polyphony you are likely to get before you run out of steam. Although Reality will run with a Pentium 90, polyphony will be rather limited — and if you have an Intel 266MHz Pentium II, this will generate four times the number of voices as the minimum recommended Pentium 133. You get the idea? Apart from the Intel Pentiums, the AMD K5 and K6 range will work, as will the new Cyrix MMX range, but not the earlier Cyrix 6x86 processors.

The Seer Systems web site has a benchmark page which shows relative performance for a range of processors. As an example, a Pentium 166MHz (non‑MMX) will use 5% overhead to play a 44kHz voice using two oscillators and a single filter. If you set up Reality to use up to 80% of processor power (see later for more details on the options), then you will manage 80/5 or 16 simultaneous voices. The same sounds using a Pentium 200 MMX will use 1.8%, giving 24 voices. The maximum polyphony available to Reality is 64 voices.

Installation

The software comes with a hardware dongle that plugs into the parallel port, and this co‑existed quite happily with those already in my machine for Steinberg's Cubase Score and the Waves Native Power Pack, although my dongle chain is beginning to bend like a banana under its own weight. The software installed quite easily, and I was soon playing away via a MIDI keyboard. Banksets contain a selection of sounds grouped together, and as well as the individual patches for each sound, these can also contain sample data. Although Reality sensibly loads up by default the Bankset that you used last session, most of the supplied ones (200Mb in total) reside on the CD‑ROM, to save hard disk space. With my ancient double‑speed CD‑ROM drive, it was 35 seconds before the screen first appeared, which was a bit tedious, especially if I wanted a different bank that session — it would be useful to be able to disable this feature.

There are some gorgeous sounds in here, and it only takes a few seconds to forget that you're actually playing a PC soundcard.

Once loaded, the software splits neatly into three main sections, each on a separate page: Bankset, Program, and Options. Bankset shows a scrolling list of available sounds in a particular bank, along with details of the algorithm type and some short text descriptions of how the sound has been constructed. The Options page allows you to set up the sounds in context. Choices include Volume, Transpose and Fine Tune, MIDI settings, Chorus and Reverb (these are global settings that apply to all current sounds), and Key response (adjusting both velocity and pressure sensitivity to suit different keyboards, actions, and sounds). Synth performance allows you to tailor Reality to your machine and application — Max CPU (20‑80%) sets the amount of your processor power that you want to devote to synthesis (if you plan to run a sequencer as well, it might be wise to wind this down a bit), while Max Polyphony (1‑64) limits the number of simultaneous voices. Reducing polyphony can result in more consistent performances if your system is tottering on the brink.

Optimise For (Faster Response or More Voices) allows you to lower the latency when playing from a keyboard, for a faster key response, or to allow more power to be diverted to playback, which is more suitable when you're employing a sequencer. The final section on the Options screen is Capture, and this is a clever way round the current main limitation of the software — that when you're running it, your normal WAV playback is inoperative. By selecting a Capture file, you can save the entire performance of Reality straight to hard disk. Although primarily intended to allow direct‑to‑disk recording, or to produce a 'sample' of vast polyphony that subsequently can be used as the basis of a huge single‑voice PCM patch, this feature would also allow you to capture an entire Reality performance. This could then be incorporated as an audio track into a MIDI+Audio sequencer such as Cakewalk Audio, Logic Audio or Cubase. Then not only will a huge chunk of processor power be released for your real‑time EQ and effects, but normal WAV playback returns, so that you can hear the rest of your audio again.

Sound Creation

The Program page is where all the really creative things happen. As always, it's easier to start with an existing patch than a blank canvas. The upper half of the page holds general data, such as program number, algorithm type, volume, pan, reverb and chorus levels, choice of monophonic/polyphonic, and control of multiple zones, where several samples have been mapped across the keyboard range. The drop‑down list of algorithms starts with PCM/Analogue, and although many of the sounds use this, it's not the cop‑out you might expect (using amazing samples to cover up weak synthesis). In fact, not only can up to four oscillators be used in a single patch, but also their 'Topology' (interconnection) can be set to one of four options. The simplest is all four in parallel, for fat analogue synthesis, but the other three are used for FM synthesis: Two Pairs (2 carriers, 2 modulators), Three into One (1 carrier, 3 parallel modulators), and All in Series (1 carrier, 3 cascaded modulators). So, samples can be used as basic sound sources, but also as modulators for FM, which adds far more scope for sound manipulation (as in Yamaha's SY series).

Nine internal waveforms are also provided: sawtooth, sawtooth2 (with some of the harmonics already rolled off), square, triangle, pulse (with full width adjustment), sine, white noise, red noise, and violet noise. Many of the analogue patches use these alone. Each of the oscillators can have its level and frequency (coarse and fine) set, or controlled from one of the available envelopes, as well as having random pitch added and the velocity response tweaked.

The other distinguishing feature of the PCM/Analogue synthesis algorithm is the provision of filters. Again, up to four can be used (one for each oscillator), although if you don't need to use a filter you can select Pass Through (no filtering at all) or No Pass (everything filtered, that is all frequencies blocked), which will both keep the processor overhead lower. Available types are low‑pass, high‑pass, band‑pass, notch, and resonator. Of these, low‑pass is the most familiar, as it's present in most analogue synths — the resonance control is lovely, really lifting the harmonics as it sweeps though its frequency range, although the Q can't be set high enough for the filter to go into self‑oscillation. The high‑pass and band‑pass options are the route to some of the more interesting sounds, and notch is simply the opposite of band‑pass, but it's the final one that intrigues the most. Resonator acts much as a band‑pass filter, but attenuation is much more rapid out of the pass band — you can send in white noise, and a breathy note emerges, rather than tuned noise. Since all synthesis is digital, stability is ensured. Thankfully, Seer have allowed wide variation on every control, although some of the gain adjustments need to be made carefully because of this, to prevent possible overload.

The Patchwork algorithm simply allows up to 16 existing patches to be combined into a single unit (Korg owners will recognise this as a Combination). Each patch can have its own volume, pan, transpose, detune, key region (for splits), velocity range (for velocity layers), reverb and chorus amount. You can put patches in the same key regions for even fatter sounds, at the expense of polyphony. The remaining algorithms use a combination of physical modelling, waveguide technology, and modal synthesis, in various combinations, and have far fewer controls to fiddle with. Each model is capable of a range of sounds, but these tend to be variations on the same theme. The most limited in range are those that are based on actual instruments (such as NylonGuitar and Clarinet), but Modal is more general, and can produce some beautiful bell‑like sounds, as well as huge drums. Tonal (using the bank of resonating filters) gives some amazing organ‑pipe sounds, although I also managed to achieve steel drums and more bells. Where specific instruments have been modelled, the controls reflect this — for example, the DistortedGuitar algorithm provides controls for Feedback and AmpDistance (which determines what frequency feeds back).

All algorithms have the use of up to four LFOs and four Envelopes, and these can be routed to many destinations. All have a wide range of options, and although I did miss having a graphic envelope with mouse‑draggable points, the problem is that there are so many permutations of re‑triggering and looping that the envelope shape plotting would be a nightmare to implement — once again, Seer have provided versatility, rather than limiting options. Finally, there's a useful selection of Reverb and Chorus settings, and although these are global (applying to all patches) the amounts can be set for each sound.

Conclusions

I've already said how much I was impressed by the sounds, and I fully expected that Reality would cost more than the AWE64 Gold soundcard needed to run it. However, at £349, this product has to be viewed in a rather different light. Far from being a clever bit of programming to give soundcard owners new sounds, Reality has the quality and scope to be seen as a cheap way of achieving state‑of‑the‑art synthesis — a sort of lateral‑thinking approach to designing a new form of synthesizer using an existing computer. Once you view it this way, and you've heard the free audio demo CD, you could be buying Reality bundled with the AWE64 Gold card and using it as a stand‑alone synthesizer. If you have access to a D/A converter in something like a DAT recorder, the resulting sound quality should stand up against that of any other modern synth, and it's more versatile than many.

Since you can load in drum sounds as samples (Seer provide 27Mb of drum loops as well), Reality could feasibly produce every sound in a complete track by itself, without any other equipment. The audio demo CD proves this — it features 27 short tracks recorded direct to CD from the digital output of Reality, using no other sources or effects. If you get the chance to hear Reality in action, don't miss it — the demo will delight Herbie Hancock fans, pianists and science fiction enthusiasts, but the real thing is even more impressive. I suspect that many people just won't be won over until they get a practical demonstration — just try to think of it as a different way to buy a synthesizer. Hearing is believing!