Digidesign's Sound Designer program gave Macintosh owners the chance to have a good look at their samples. Now the company's latest program, SoftSynth, offers a new way of creating samples from scratch - without recording from the 'real world'.

It's 1986. People (not all of them musicians) are buying samplers - machines that use computer technology to record sounds and reproduce them in a musical way. The samplers, in turn, are becoming available in many different shapes and sizes, and at a variety of price levels. But sad to say, these samplers have so far tended to sound much the same as each other.

This has been less the fault of the samplers themselves, and more the responsibility of the people who've been using them. Or maybe the real culprits are current trends in popular music, which dictate that musicians and producers come up with no sound whatsoever that hasn't been heard on the radio a hundred times before.

That's ironic when you consider that samplers can put a whole world of new sounds at musicians' disposal, even though they have a habit of imposing their own characteristics on the noises they record and reproduce.

To get away from all this, wouldn't it be a good idea if we could use the same computer technology to create sounds that fall outside the realms of 'samples' in the accepted sense?

Well, the Digidesign SoftSynth program is a clever way of getting a sampler not to sound like a sampler. OK, so there's a degree to which a sampler will always operate like a sampler, churning out data at various speeds with 'chipmunk' effects when samples are played at pitches much higher than the original sample, and the 'grunge' of clock noise when samples are played far below.

But at least you can do something about the sample source material. It may appear that samplers are limited to manipulating 'real' sounds, but SoftSynth proves that a computer-based system can be used to circumvent the sampling process altogether, and produce truly original sounds.

What SoftSynth does is create samples on the Apple Macintosh, rather than merely edit samples which come from the 'real world'. These sounds are created by additive synthesis, and in the case of SoftSynth, this means it's possible to control the mix of 32 harmonics through the duration of a sample. If this were to be implemented in analogue hardware, it would be the equivalent of having 32 oscillators, each with its own complex envelope. Even if digitally-generated oscillators were used, the hardware required for such a synthesiser would still price it far beyond most musicians' reach.

Instead, SoftSynth analyses the harmonics you tailor with Mac's high-resolution graphics, then calculates a sample which can be downloaded for playback by a sampler. A number of samplers implementing the MIDI Common Sample Dump Standard can receive a SoftSynth sample dump, or you can choose to create a disk file compatible with Digidesign's own Sound Designer software package.

Since the Mac is not required to spend all of its time generating sound as the SoftSynth controls are adjusted, the controls themselves are easy to use and take advantage of the Mac's graphics whenever possible.

SoftSynth's main programming screen shows a frequency analysis of the current sound, with all 32 harmonics clearly displayed. Below the frequency analysis are 32 'faders' corresponding to the harmonics. If you want to change the harmonic mix, simply 'grab' the desired fader with the Mac's mouse, and move it. The frequency analysis doesn't respond to such changes instantly, though. To update the display, grab the eye icon, and after a moment, the display features the edited harmonics.

The frequency analysis shows the way each harmonic's amplitude varies in time, but there's more to each harmonic than can be seen from this screen. Grabbing the number of the desired partial from below the faders takes you into Single Partial editing mode. The screen shows a section of the selected harmonic's amplitude envelope, and below that, its pitch envelope. Yes, each harmonic can be detuned by a complex envelope - and by an initial amount as well as by the pitch envelope.

The envelopes themselves are certainly more flexible and easier to adjust than those provided on most synthesisers. Each can have as many as 40 linear segments, and is shaped by simply grabbing any segment with the mouse, and then stretching it into the desired shape. The usual Macintosh mouse-activated scrolling controls allow you to look closely at selected portions of the envelope.

This part of SoftSynth also incorporates a feature that's been sadly lacking on many instruments - envelope copying. This avoids the drudgery of creating envelopes from scratch every time you build a new sound, and it's just as well the program has it, when you consider the complexity of SoftSynth's envelopes.

And since each envelope is easily edited, it's a simple matter to generate 32 variations on one envelope for the harmonics, examine the frequency analysis as you audition the sound, and then continue editing as you deem necessary. Incidentally, you can listen to each harmonic individually if you want to.

As if these features weren't enough, each harmonic can be a sine, sawtooth, or square waveform, or any of three band-limited noise waveforms. This is a deviation from standard additive synthesis, which is based on Fourier analysis and hence on the combination of sine waves - but so what? Things are more versatile this way. The band-limited noise is a nice alternative to the white noise usually found on analogue synths, too.

With an understanding of Single Partial mode, a whole new realm of sound is open for exploration, though you do need to be patient.

But that's hardly where SoftSynth stops. An unusual icon showing a knife cutting into an alarm clock is the gateway to Time Slice editing mode, in which all 32 harmonics are adjusted simultaneously instead of one at a time.

The Time Slice editing screen shows a master envelope which controls the amplitude of the sample, while numbered rectangles (representing timbres) are displayed along the envelope's time axis, coinciding with break points in each harmonic's amplitude envelope. Since there are 40 potential break points in each harmonic's envelope, and there are 32 harmonics, there could be a lot of timbre break points appearing in Time Slice mode. To avoid confusion, SoftSynth eliminates the least significant timbre break points, and displays only the most significant ones.

These timbres can then be moved around, or copied to other parts of the sample by grabbing the desired timbre with the mouse, then dragging it to another point on the master envelope. This is a 'brute force' method compared to Single Partial editing, but it's extremely powerful.

As it turns out, these two editing modes are actually interactive. That is, you can switch between Single Partial and Time Slice editing modes at will. A big plus.

The range of sounds SoftSynth can create is pretty much limited by the amount of time you make available - particularly as it takes a little while for the program to calculate a sample from the harmonic structure you've come up with. And as with any synth, it's handy to have some presets to use as starting points while you learn how the system works. In response to this, SoftSynth will come with a palette of 15 (or so) preset harmonic files, so you can start with a piano or brass sound, then alter it as you please.

One feature which had not yet been implemented, let alone announced, at the time we examined SoftSynth is a 'random' sound generator. Sounds silly, I know, but as sound-creation systems become more complex, such features are bound to become more popular. Sequential's Prophet VS features a similar function as part of its voicing architecture, while DX editing programs from Hybrid Arts (the DX Droid) and Steinberg Research (the Pro Creator) employ the same sort of approach - though it should be stressed that there's little resemblance between the results of the VS, DX and SoftSynth systems.

Apparently, the random option of SoftSynth will let you set limits on certain parameters, so that you can loosely define the type of sound you want to produce. The program will then generate random harmonics based on these restraints. Should be interesting.

The final stage is loading your synthesised sound into a sampler. The 'Host' (or 'Sampler') menu lets you set up the Mac for the MIDI interface in use. Following that, the 'File to Synth' function (selected from the 'File' menu) takes care of sending the sample data to the MIDI-equipped sampler in question. The sampler can now play the sound no differently than it would have if you'd sampled it from a rather more expensive instrument. The big difference is that you made up the sound yourself, and that the SoftSynth program costs less than any samplers with 'professional' aspirations.

Summing up, it strikes me that SoftSynth could be something of a dream come true for many synthesists. The controls are there for endless tweaking of sounds, and if inspiration or your patience wanes, you can always resort to generating random samples. Your sampler will never know the difference.

Price To be announced

More from (Contact Details)