Computer music systems seem to be dividing off pretty cleanly into two camps at present: those that use additional hardware (analogue or digital) for waveform synthesis (e.g. MH MusicSystem, Fairlight CMI, PPG Wave 2, etc.); and those where synthesis is totally software driven with no additional hardware other than a D/A converter (e.g. MTU Instrument Synthesis).

The Soundchaser Computer Music System falls into the first category, but, unlike the majority of its competitors, uses analogue voice cards for synthesis. As a result, the synthesis of the Soundchaser is inherently limited to hardware-fixed waveforms, but what's lost on the swings is gained on the roundabouts, as the analogue voice card approach allows filter sweeps to be easily implemented. Such filter sweeps can also be executed by a system using wholly digital means of synthesis, but this involves multiplying or dividing the contents of a wave-form table before it's directed to the D/A converter, and this takes too much time (100 to 150 microseconds for an 8-bit by 8-bit multiplication) to be really practicable for real-time music synthesis with present-day microprocessors. Perhaps the hundred-times faster Josephson computer will change all that and provide you with a deep, deep-freeze at the same time!

Digital systems therefore opt for the more palatable alternative of switching the D/A converter input between different waveform tables, each of which may correspond to a different harmonic complex, thereby producing dynamic waveform changes or 'timbral sequencing'.

In practice, it's not actually that easy to digitally reproduce the soaring filter sweep so beloved of the present generation of 'futurist' synthesists, even with an instrument like the Fairlight CMI, although, in theory, the sampling of a bog-standard analogue synthesiser doing its VCF thing via a fast A/D converter should provide the instrument with suitable fodder for it to manufacture something pretty close to it.

What this means for digital synthesis systems lacking the sophistication of fast sampling techniques, or multiple waveform table entries for each voice, is that the sounds produced lack the hallmark of a Moog-type VCF that's more-or-less de rigueur for pop musicians concerned with imitating each other's sounds and styles. As supply and demand is the name of the game, it makes good sense for a computer music system like the Soundchaser to aim for the best of both worlds by using digital control of largely analogue circuitry.

The System

The Soundchaser is configured around an Apple II with 48K RAM and Applesoft BASIC, a disk drive, and a monochrome monitor. Additional hardware consists of a four-octave keyboard, with the standard type of diode encoder, sending key-down data to 4051 analogue multiplexers on an interface card. This plugs into an expansion connector on the Apple II motherboard along with one or two three-voice cards.

Each voice on the card uses standard chips and circuits for the 24 dB/octave low-pass filter (CEM 3320 VCF) and VCA (LM3080 transconductance op-amp), but, rather than following this traditional analogue approach through with a multi-function VCO like the CEM 3340, Passport Designs have elected to use an Intel 8253 for the purpose of a primary waveform source. The 8253 contains three 16-bit timers (one for each voice) which are used to count down from the 2 MHz Apple Q3 system clock in accordance with a binary number output along the data lines, thereby producing a square wave of a particular frequency. Obviously, the larger the reference frequency and the number of bits being used for counting down, the greater the accuracy of high frequencies. The Soundchaser programmable 'oscillators' perform well in this respect and achieve the value of 6649 Hz for 8th octave G sharp rather than the spot-on equal tempered value of 6645 Hz. The hardware picture is completed by a 'wave shaper' (ramp generator), which turns the 8253 square wave outputs into sawtooth waveforms, and a multiplexed D/A converter for control of the VCFs and VCAs.

The preliminary Music Operating Software comprises two principal programs: EDIT, which allows the musician to preprogram the oscillators, VCFs and VCAs of the voice cards by using an interactive display to enter parameters in the form of 'soft' switches, sliders and contours; and the SEQUENCER, which assigns record or playback modes of four independent multi-event banks holding presets and polyphonic sequence data entered from the four-octave keyboard.

Edit

The Soundchaser EDIT program and display (Figure 1) are accessed via the main menu (Figure 2). EDIT uses conventions of analogue synthesisers in programming the various parameters of the voice card processor chain, but with distinctive features that are both good and bad.

One really powerful feature is the ability to draw four contours on the screen (using game controls) and then use these to control the various sections of the voice cards. Figure 3 illustrates the patching options for the modules and the controls that can be assigned to them. As these contours can, in theory, be any size, shape, or form (in fact, Passport advise the user to think of the display as a 'space in which to draw your ideas'), it does open up remarkable possibilities for creating complex sounds.

Whilst it is feasible to draw and use a complex envelope or LFO waveform, the fact that each contour is actually read by software as 64 points, rather than a continuous function, means that such artistic extravagances have a habit of producing choppy sounds. Drawing the sounds is fairly straightforward, but the 'small, blinking cursor', used for locating the contour points on the bottom line of the top half of the display, is very small indeed, which doesn't help matters. Furthermore, changing just one aspect of an envelope (say, the gradient of the attack portion) necessitates re-drawing the entire envelope!

These aren't the only limitations to one's synthetic aspirations. Having drawn in some likely contours, the bottom half of the EDIT display is engaged by entering 'VP' (Voice Panel). This allows the user to set the 'soft' switches and sliders with which to apply the contours to the voice card operation. Once this has been done, 'P' (Play) can be entered and the keyboard actually played.

But where aspirations really fall badly down the hill is in the realisation that all these voice card changes have to be entered in non-real time, so that, unlike a conventional analogue synthesiser, it's impossible to change knob settings or patches (i.e., the 'soft' switches, sliders or contours) as one's playing, a curious deficiency for a synthesiser proclaiming itself to be capable of 'providing a very wide range of sounds and effects unsurpassed by computer synthesisers costing $5000 and up'.

Another point to bear in mind is that each voice is derived from only a single oscillator, with none of the facilities normally used in synthesisers to 'thicken' this sort of voice, such as pulse-width modulation (PWM) or the mixing of the sub-octave and other waveforms to the basic oscillator output. Whilst it's true that a sawtooth waveform is also available, courtesy of the wave shaper, it's only as an either/or option, and that's as a result of fiddling about with a DIP switch on the voice cards - inside the Apple - hardly a real-time control!

To be fair, Passport Designs tell me that they do intend implementing some form of real-time control of the voice card parameters in the near future, and also plan to bring out a dual two oscillator/voice card using Curtis chips throughout for VCOs, VCFs and VCAs. This should provide a range of sounds equivalent to those obtained from a processor-controlled synthesiser like the Prophet 5, which also uses Curtis chips. Mind you, I do think it's a little imprudent to launch a system like the Soundchaser with voice cards that are likely to be upgraded in the near future to something nearer the real McCoy - especially when each three-voice card retails for $350 (probably £250 here) - perhaps Passport would consider a trade-in of the old cards for the new?

Sequencer

According to Passport, 'the Sequencer is an extraordinarily flexible and expressive sub system which expands your role from a performer/programmer to an arranger/conductor'. The SEQUENCER display in Figure 4 shows that four banks of presets (instruments programmed with the EDIT program) and sequences (notes or chords played on the keyboard) are available for arranging (setting a preset to a sequence), mode assignment (record, playback or off), and tempo variation (1 - fastest to 16 - slowest). Though the SEQUENCER will store a fairly long monophonic line if the tempo is slow, Passport are tempting providence by stating that each bank is 'capable of storing 2 to 16 minutes of music', as each of these is limited to only 256 events. This isn't exactly generous by contemporary standards of polyphonic sequencers, but the future is likely to see this extended to something like 16K-worth of storage.

The system allows the keyboard to be played on top of a sequence playback (up to the limit of available notes), and it's also possible to program the real-time playing with a different voice to that of the sequence playback. When playing on top of a sequence the keyboard always takes priority for note assignments, although, if one does try to play with more than one's fair share of notes, strange things start to happen to the sequencer timing, i.e., parts get inexplicably out of sync. Actually, this is the easiest way I've ever found of setting up overlapping loops - it's a shame it's all accidental!

Onwards, Ever Onwards

The basic sound quality is good, even though some quantisation noise creeps in via the timer chips (a further section of the DIP switch applies a 10kHz cut-off to remove this), and the range of synthetic possibilities is considerable owing to the 'etch-a-sketch' contours.

In the States the total system sells for $1350 ($650 for the keyboard and software, $350 for each of the voice cards), but, over here, I don't see much chance of the price being less than £1,000.

The future is likely to see much greater expansion of the Soundchaser's capabilities, with 'professional series software' for teaching, and 'musician series software' offering more advanced musical utilities such as instant recall of presets and layered sequencing. Also, a Music Sketcher offering real-time transcription onto the monitor screen of monophonic lines played on the keyboard, together with subsequent editing of the entered music, is under development and likely to retail for around $100.

Another item that Passport are developing is a software package for running the Mountain Hardware MusicSystem boards from their own keyboard. They hope to sell this for £650, which will put them in direct competition with the Alpha Syntauri system selling for more than double that amount in the States.

Passport Designs obviously have plenty of bright ideas and some considerable programming ingenuity, but these qualities only make it more difficult to understand why the Soundchaser has been released with limitations that can only serve to frustrate musicians being introduced to computer music systems for the first time.