What can you do with a sampler when you're bored of sampling? Paul Ireson finds an answer to this question in the form of creative software from Digidesign that can turn your sampler into a powerful synthesizer.

The idea of unlimited sonic horizons is manna to most electronic musicians. A machine that could actually offer such possibilities - rather than just claim to do so, as many have - would be a visitation of heaven on earth to these people, and I'd count myself amongst the crowd who are awaiting the arrival of their sonic messiah (a Prophet perhaps?). Of all the instruments that technology has so far offered us, only the sampler really approaches this flexibility, through its ability to reproduce any sound that we sample. But it lets us down in one important respect: it can only reproduce existing sounds. Samplers are therefore not inherently creative sonic tools. (I'm well aware that some samplers offer extensive editing facilities, and that using sounds out of context can be creative, but bear with me; I'm trying to make a point...) It is to this shortcoming that an American company called Digidesign have successfully addressed themselves, by producing software that takes the range of sounds that a sampler can create into totally new territory. The programs in question are Softsynth and Turbosynth, which effectively use a sampler as the output stage of powerful synthesizers.

These 'synthesizers' exist only in the form of software simulations, through which the user has the impression of manipulating real synthesis elements on the computer. These elements - which could be oscillators, filter and amplitude envelopes, etc - relate only to parameters in the computer's memory rather than 'real' hardware, but they can be heard when the computer transfers the 'virtual' sound to a sampler (entirely in the digital domain, via MIDI), and that's real enough for me. It's important to emphasise that both programs are not synthesizers in the sense that they create sounds in real time: rather, what they do is to create single samples to download into a sampler, which then becomes the medium for real-time playing of the sounds.

Neither Softsynth nor Turbosynth are new - both have been available on the Macintosh for some time, and have now been ported over to the Atari ST, Nevertheless, it's worth reporting on them even now, as both programs can give a whole new lease of life to a sampler, and no-one else is producing anything quite like them. Turbosynth has already been reviewed in Sound On Sound, so this review will focus on Softsynth (but see panel). The hardware requirements for running Softsynth are a 520 or 1040 Atari ST with mono or colour monitor, and a compatible sampler. I used my old Akai S612 without any problems - I wanted to use an S1000 as well, but unfortunately the version of the program supplied for review (2.0) does not yet support that machine.

With the appropriate hardware and the Softsynth program, you effectively have at your fingertips a synthesizer that can use up to 32 Partials (oscillators) in a combination of additive and FM synthesis, with full control over complex amplitude and pitch envelopes for each of these Partials. The combination of additive and FM synthesis proves to be a good one, for the two techniques complement one another rather well: whilst additive synthesis allows considerable detail to be added to a sound, FM allows the painter in sound to work with a larger brush, as it were.

OPERATION

After loading, Softsynth presents the user with its main screen, from which the large-scale editing and all of the sound transfer operations needed to move a sound from the ST to a sampler can be carried out. Most of this screen is occupied by a graphic representation of the current sound file in the form of a 'mountain range' harmonic display - time is displayed on the horizontal axis, and the sound length indicated in seconds on that axis. Along the bottom of the screen are 32 faders which control the overall levels of the 32 Partials that Softsynth uses to create its sounds. To the left of the mountain range are displayed three of the most fundamental sound parameters: Frequency (which determines the fundamental frequency of the sound being created), Sample Rate (which is really only important when transferring the sound to a sampler) and Length (in number of samples). All three can be changed at any time. More detailed editing than the main screen allows is carried out by operating on single Partials, or with the Time Slice editing mode. More about these features later, but in the meantime let's concentrate on the basics of operation.

Whenever you want to hear the sound on which you are currently working, you have two options: you can either 'preview' it on the ST's internal speaker, or transfer it to your sampler to be heard in its 'proper' form. Both procedures require the ST to spend a little time 'synthesizing', that is actually creating the sound (or at least the sample data for it) that the sound parameters describe. In order to preview a sound, you click once on the loudspeaker icon on the main screen. The ST then spends a few seconds thinking things over before the sound is played on the ST's speaker. Provided no changes are subsequently made to the sound, you can hear it again without having to wait by clicking once more on the speaker icon. If the sound is to be transferred to a sampler, it must first be synthesized then saved to disk as a Sound Designer sample file — the Synthesize option from the File menu achieves this. That file must then be transferred via MIDI to a sampler, using the File>Sampler function.

Before attempting to transfer the sound you must select the correct option on the Sampler menu, to specify the model of sampler you are using. The program automatically selects a suitable default sample rate whenever you specify a new sampler, but if the chosen sampler has more than one possible sample rate (which it probably will) you may have to change Softsynth's Sample Rate parameter to match that of your sampler.

The synthesizing process, either for previewing or prior to transferring sounds, can unfortunately be rather time-consuming, and the saving to disk and data transferral add even more waiting time. For example, a sound that took just over 10 seconds to synthesize for a preview required 25 seconds to synthesize and save to disk, and 17 seconds to transfer via MIDI to a sampler. The sound itself was just over a second long. The upper limit on sound length seems to be 37.5 seconds, provided your sampler can handle that much data. I used a RAM disk to speed up the disk saving and loading times, but this still leaves the synthesizing and transfer times unaltered.

Whilst the Sound Designer sample files are necessary as an intermediate stage to transferring sounds to a sampler, Softsynth's sounds can also be stored in a different and more economical way: as parameter files. These simply contain the data necessary for Softsynth to reconstruct the configuration of all 32 Partials that contribute to the sound.

ADDITIVE SYNTHESIS

Softsynth's primary method of creating new sounds is through additive synthesis - ie. combining the outputs of a large number of simple oscillators, with no filtering of any kind. The FM synthesis capability was added in one of the program's early revisions. Additive synthesis is perhaps the most basic and yet most powerful synthesis technique available. Since any sound can be broken down into a combination of simple sine waves, of different (and possibly time-variant) frequencies and amplitudes, it follows that if we combine enough sine wave oscillators and have complete control over their frequencies and amplitudes, we can create or re-create any sound at all.

Softsynth provides 32 Partials (oscillators) for you to play with, and the timbre (harmonic content) of the sound at any point is determined by the relative levels, frequencies and waveforms of the Partials. Basic amplitude levels are set on the main screen, and the frequencies default to a straightforward series of whole number multiples of the fundamental frequency, from 1 to 32. To carry out more detailed editing of each Partial individually, a Single Partial edit screen can be called up for any Partial by simply clicking on its number, shown below its fader on the main screen.

The two most important Partial parameters, both shown and editable within the Single Partial screen, are waveform and frequency. Although additive synthesis allows the harmonics of a sound to be created by stacking up numerous sine waves, the facility for Partials to use other waveforms provides a very useful shortcut in many cases, at the expense of losing the precise degree of control that using only sine waves gives. The wave shapes available are sine, triangle, square, band-limited noise, and white noise. The two noise waveforms are useful for creating effects such as the breathy chiff at the start of a flute sound.

The Partial frequency is expressed as a ratio of the fundamental frequency, to a resolution of two decimal places. As a rule of thumb, whole numbers provide pleasant, musical harmonics, whilst anything in between (1.37, 13.8 etc) produces a dissonant, metallic effect. The screen provides graphic displays of both the pitch and amplitude envelopes for the Partial, and allow straightforward click-and-drag editing of the envelope shapes. The amplitude envelope can include up to 40 stages, and the pitch envelope up to 15. The range of pitch variation produced by the pitch envelope can be switched between 2% and 50% (for each Partial individually): the former range is generally useful for producing subtle modulation and movement in a sound, the latter for 'special effects'.

Whilst single Partial editing allows the sound to be edited in very fine detail, in many ways it is too fine, and is not suitable for making overall timbral changes quickly. Time Slice editing allows you to do just that. The Time Slice window (called up by clicking on the butcher's knife/clock icon on the main screen) allows a total amplitude envelope for the sound to be edited, avoiding the necessity to alter every Partial's attack time individually if you want a slower attack, for example, and perhaps more usefully allows the relative levels of all Partials to be examined and altered at specific points in time. Once new levels are set, Softsynth will interpolate between the new values and the adjacent timbre events. As is intended, Time Slice editing provides an easily-grasped means of controlling the time-variant timbral characteristics of a sound.

A second means of altering sounds on a large scale is provided by the Smartsynth function. This allows you to create a 'template' which describes a sound, in which one of three options is chosen for whichever of the sound characteristics you choose to use. The characteristics include things like Attack, Harmonic Range, Doubling and Sustain level, and the three options might be Slow, Medium and Fast (for Attack). Within the limits allowed by each active characteristic, Smartsynth will generate random values for all the relevant parameters. The function can be used either to create entirely new sounds, perhaps as a rough starting point for something else, or to modify an existing sound. In this latter application, the fact that only parameters relating to 'active' characteristics are changed is obviously important. I used the function mostly for its ability to create a doubled sound - it will copy Partials 1-16 over to 17-32 with slight detuning applied, producing the sound of two detuned sources - and also to do simple things, like reducing the higher harmonics in a sound that was too bright.

FM SYNTHESIS

Additive synthesis is characterised by the need to use many Partials to produce complex sounds. Softsynth's ability to use FM synthesis as well is therefore a good complement to additive, because it can create complex timbres with only a few Partials. The FM employed here is, in principle, the same as that on Yamaha DX and TX synthesizers, though in one or two significant respects Softsynth is more powerful. Whereas DX and TX synths only have four or six operators, and a small number of algorithms, you have almost complete freedom to use Softsynth's 32 Partials as modulators and carriers.

The program's FM patching screen allows the user to assign any Partial to modulate any other, subject only to the restriction that any Partial can only be modulated by one Partial. Otherwise, anything goes: one Partial can modulate 1,2, 7 or 31 others, or even all Partials including itself; a huge stack of all 32 Partials could be created, in which each modulator is in turn modulated by the previous Partial... or anything in between. I tried to work out just how many 'algorithms' are possible here, but my rusty applied maths just wasn't up to it.

Another major difference to Yamaha FM synthesis is that for each of the modulators, you can specify whether it is also to be heard as an audible Partial in its own right, or whether it is to be removed from the mix and only heard through its effect as a carrier.

CONCLUSION

It's not hard to make an evaluation of Softsynth's usefulness: it is an almost unique piece of software that can give you access to an entirely new sound palette. As such it is an exciting program to use, and cheap at the price. However, it is not a program to be taken too lightly - if you don't have the time to get to know the program, and are not likely to do enough sound programming, the combination of the detail required to create good additive sounds and the accumulated delays of constant synthesizing of sounds could make the program more trouble than it's really worth. But taking this into account, if you feel you could get into a new way of making sound or are just looking for ways to breathe new life into your sampler, Softsynth is most certainly for you.

FURTHER INFORMATION

£235 inc VAT.

Sound Technology plc, (Contact Details).

We have already reviewed Turbosynth in its original form for the Apple Mac [see SOS December 1988], but now that it's also available for the Atari ST, it is certainly worth another mention. The program is unchanged in all but minor detail, and therefore becomes to the world of ST software what the original version is to that of the Mac: a uniquely powerful and exciting sound synthesis tool.

Turbosynth is quite different from Softsynth in its synthesis technique, and is undoubtedly more versatile. It is essentially a software simulation of a modular synthesizer, the like of which has never actually existed: the program's modules are selected and edited on screen, and you can use as many of these modules as the computer's memory will allow (about 10-15). The original sound sources are Oscillator and Sample playback modules, and the modules that can be used to process the basic signals are: Amplitude and Filter Envelopes; a Spectral Inverter; Delay module; Pitch Shifter; Modulator (offering Pitch and Amplitude modulation); Waveshaper; Stretcher; Resonator; and Mixer (which can accept up to 32 inputs). The virtual audio signals are routed between the modules with onscreen patch cords.

I refer you to the original review for details, but in this limited space, suffice it to say that Turbosynth allows the combining, editing and complex processing of both synth-type waveforms and sound samples in a program that is as easy to use as it is versatile. Its only significant drawback is one it shares with Softsynth - namely, the length of time that is required for the synthesizing and transferral of its wonderful sounds to a sampler.