Making More Of The Kawai K5
Martin Russ reflects on a cult phenomenon: Kawai’s K5 additive synthesizer.
Before the K1, Kawai's presence in the hi-tech market place was limited. A reputation for high quality, high specification instruments had gradually been building with the K3 synthesizer and R100 drum machine, but it was the K5 additive synthesizer which finally showed that Kawai were deadly serious.
The K5 came out in 1987 — at almost exactly the wrong time! The original Yamaha DX7 was beginning to fade and had been replaced by the improved Mark II variant, Roland's D50 was virtually an industry standard for pro keyboard players, and the last thing most people wanted was an instrument with almost four times as many parameters as a DX7, a very complex and flexible synthesis method, no samples in ROM, no on-board effects, and no fashionable sounds. You might think that the K5 would fade into obscurity against competition like that, but there will always be those hardened individuals who search out the unusual and obscure, in a constant quest for the hidden gem.
As if to help such people, Kawai provided an inducement — a bargain price. Last year the K5 was sold off, worldwide, at a quarter of the original retail price! Not surprisingly, it sold well (and very quickly!), mainly to those who wanted a low-cost sound source, but also to some people like myself who recognised a bargain when they saw one! At the price, the K5 represented the 'Bargain Of The Century.' Why?
The K5 is a remarkable piece of engineering. It uses a completely different method of synthesis to almost every other sub-£2000 instrument ever produced, namely additive synthesis. Normally restricted to serious researchers and large mainframe or minicomputer-based electronic music facilities in Universities or Colleges, the process of adding together individual harmonics to create sounds is incredibly powerful, but previously prohibitively expensive. At its full retail price, the K5 had (and probably still has) arguably the widest range of possible timbres of any instrument in its class, and at the lower price it has no competition at all.
Additive synthesis uses lots of separate sine waves at different frequencies, with lots of different envelopes to control them, either in groups or individually. The output is the sum of all the sine waves — rather like Algorithm 32 on a DX7, or the drawbars of an electronic organ. This trivialises the possibilities because, unlike either a DX7 or a drawbar organ, the K5 can deal with many more harmonically related sine waves, and offers more envelope control possibilities.
The K5 adds a few extra features which make additive synthesis a little easier to deal with. Grouping of harmonics to envelopes keeps the number of envelope generators manageable, and a real-time digital resonant filter (a digital VCF!) helps provide coarser overall control of timbre in an 'analogue' way. What this means in practice is that you have a synthesizer with two detunable oscillators which can use any waveshape you like, built up from up to 126 sine wave harmonics. This waveform is not static either, because the harmonics can be controlled with envelopes to produce dynamic waveform changes. More interestingly, using a computer it is possible to convert samples into harmonic information and use this to provide a resynthesis capability, although few people seem to have exploited such facilities!
Regardless of how you get the raw sound, it can then be filtered by the equivalent of a sophisticated VCF to give broad timbral changes, passed through a digital VCA, and then a graphic equaliser. Up to 15 different sounds can be stacked (the K5 is multitimbral also) to provide humungously big and fat sounds, whilst the four separate audio outputs provide scope for complex stereo imaging effects. Faced with synthesis power like this, the K5 was an essential purchase for a serious programmer like myself. And it is this which has helped the K5 acquire a cult following amongst informed synthesists.
There is always a catch to any bargain. In the case of the K5 the factory presets were OK, but not quite in the same class as the D50's 'Atmosphere' or 'OK Chorale' etc. Most importantly, few third party companies seemed to be programming sounds for them. A few banks of sounds were available from Kawai, but these generally fail to exploit the power lurking inside the instrument. The few excellent voices which are included begin to show just what the K5 can do, although the best advert is possibly the majority of Jan Hammer's music for Miami Vice, which utilises plenty of K5 sounds [see interview: SOS Jan 1988].
Editing software and utilities for the K5 seem to be few and far between, with only Dr.T's K5 editor for the Atari ST being a notable contender — and a very good program it is, too! Even so, programming all the parameters in a K5 can be slow work — in many ways it is a return to the old 'analogue' style of synth voice creation. With over 500 parameters (more than three times as many as on a DX7!) to investigate, it is a good thing that the filter and the graphic equaliser are there to make rapid tonal changes, but the scope of full programming with the K5 is enormous.
Faced with a lack of decent sounds and a time-consuming programming method, many purchasers of the K5 have probably not attempted to exploit its potential. Since I seem to write software which nobody else does, the time seemed right to create a utility program for the K5, to help distribute some of the better K5 voices amongst users, and also to provide some useful extras to make programming the instrument a little easier. The result of several months investigation and programming is another SOS Software program: the K5 Utility.
The K5 Utility is a Public Domain program which acts as a librarian for Single sounds (not Multis at present) and for Harmonics. Quick access to standard 'analogue' waveforms in the form of harmonic information can be sent to the K5, and there is even something like the equivalent of the PCM samples found in the Roland D50 and the Korg M1, except that these can be loaded and saved to floppy disk.
As with most of the programs in my 'System Exclusive' series, the K5 Utility is only available for the Atari ST, and will only run in monochrome. The K5 Utility is available from SOS Software — see end of article for details.
In the course of using my K5, I have developed a few useful techniques which may prove helpful to other K5 owners. The complexity and sheer number of parameters make the K5 a daunting programming proposition, but the end results usually justify the effort.
Investigate the 'twin' and 'full' options. Most of the factory voices use the 'twin' setting to provide two parallel sets of 63 harmonics, giving you the equivalent of two separate sounds at once in 16-note polyphony. 'Full' mode only offers one sound but gives you a single set of 126 harmonics — Section 1 (Kawai always refer to the two Sections as S1 and S2) provides harmonics 1 to 63 (the 'lower' ones), whilst S2 provides 65 to 127 (the 'upper' ones), with harmonic 64 being lost in the development rush, apparently!
At first glance, these extra harmonics do not seem to gain you any advantage, since the only time that you will hear the upper harmonics (65-127) will be on very low bass notes — see Figure 1. But the real use of 'full' mode lies in retuning the upper harmonics down, so that they fall in the same frequency range as the lower ones. When you do this the spacing between the harmonics shrinks accordingly, with the result that you obtain a very useful set of harmonics for 'filling the gaps' left between the standard 1 to 63 lower harmonics.
If you take a 'twin' sound and convert it to 'full', then transpose S2 down to -36 or -48, you can hear the characteristic detuned, bubbly sound which results from having lots of harmonics at unusual intervals. In fact, when you move the upper harmonics down in frequency in this way, you end up with a single octave span, but with 63 harmonics covering it. Figure 2 shows this in diagrammatic form.
When you are using the DHG pages with the upper harmonics, you will discover that the Octave and 5th selectors do not actually affect any harmonics — for the Octaves this makes sense, since the octaves are at harmonic 64 and 128, and are thus outside the available harmonics, but the 5th should be active — this looks like either an oversight on Kawai's part or a genuine software bug!
The other interesting phenomenon which occurs is also found on Yamaha's DX synths — if you set the frequency too low, then the notes repeat themselves on an octave basis. It is possible to set up sounds which play over just a two octave range!
If you set lots of the harmonics to high values, this creates a useful noiselike sound which is suitable for all sorts of digital and sample-like sounds when it is filtered or mixed with a sound from S1. Setting just a few harmonics results in 'beating' tones, because of the unusual relationship between the harmonics.
These sounds can be used to add interest and unusual touches to otherwise ordinary sounds, and can often disguise the identity of the K5. Remember that in 'full' mode you still have S1 with all the usual facilities, just as normal — so you can combine S2-based sounds with ordinary S1 sounds as you wish.
By setting the keyboard tracking parameter (KEY) on the DFG page to KEY and setting a suitable note, you can fix the frequency of S2, and so provide sounds which do not transpose across the keyboard range. This is useful for breathy chiffs, which only require slight additional filtering to provide suitable character changes across the keyboard range. For very low notes, the character of the S2 'full' harmonics becomes very metallic and cyclic.
In 'twin' mode, the K5 Utility program can be used as a quick way of accessing classic waveshapes like square, sawtooth, rectangular, etc. By using these 'preset' harmonics to produce the basic sound, the DDF can then be used, in exactly the same way as a VCF in analogue synthesis, to produce the filter sweep and brassy sounds which were so popular in the Seventies.
The K5 has independent pitch control over the two Sections. This makes the creation of brassy slurs and string pulls easy, especially since they can be restricted to just one Section — the other can provide a true pitch. But there is rather more to using pitch envelopes than that...
By setting all the levels to -31, and the ENV depth to a high number, you can produce very low frequencies. Because the start and end pitch is preset to zero, you will always get some audible pitch change unless you set Rate 1 to zero — you cannot alter the start and stop pitch, unlike the pitch envelope on 6-operator Yamaha DXs. You will also need to be careful to ensure the pitching is correct — if you set the envelope levels to -31, then the values shown in the ENV depth parameter on the DFG page are in semitones — so 24 will move the sound down by two octaves. Remember that the octave repeating starts to operate at low frequencies.
The pitch envelopes can be looped around their sustain segments, which opens up the possibility of vibrato and pitch blurs — virtually the equivalent of two extra independent LFOs! Rates 3 and 4 set the speed and the duty cycle of the looping section, and Levels 3 and 4 set the maximum and minimum excursions of the pitch deviation. The DFG ENV parameter sets the overall depth in semitones (when Levels 3 and 4 are set to +/-31). Note that the levels affect the speed as well, since it takes longer to make larger changes.
Obviously fast, wide pitch changes sound awful, but two effects are very useful: vibrato with asymmetric duty cycles can improve string sounds amazingly, and pitch blurring can make the most synthetic sound take on a fascinating 'organic' character.
|Rate 3:||4 or 5|
|Rate 4:||4 or 5|
|Level 3:||+15 (approx half semitone)|
|Level 4:||-15 (approx half semitone)|
|DFG ENV:||1 (about +/-1 semitone overall)|
So far we have produced some unusual sounds and pitch modulated them. The DDF (filter) is usually the next port of call, because it is quick and very easy to use subtractive techniques to produce classic 'analogue' filter sweep sounds. An alternative approach is to turn the DDF and DDA off and let the DHG straight through to the output, using the envelopes for the DHG to control the timbre changes. This is real additive synthesis, instead of subtractive filter-based synthesis, although you can always turn the filter on after setting up the DHG envelopes...
Four envelopes assigned to 63 harmonics can be confusing. I tend to work out groups of harmonics using both DHG pages to enable the harmonics associated with envelopes to be turned on and off, so that I can see and hear what they sound like. I usually choose the basic underlying tone first and use Envelope 1 to control it — typically this is the first sound to appear, and so it has the shortest attack time. The rest of the harmonics are assigned to envelopes in order of importance, so that the four envelopes are most efficiently used. If any harmonic is needed both in the basic and another group, then I either split the groups of harmonics into two, or alter the basic envelope.
There really is no substitute for detailed listening and exploration of any available sounds. The more familiar you are with the structure of sounds, especially their harmonic structure, the better. Dr.T's K5 Editor has a wonderful 3D display which shows how the harmonics change with time, and this can help enormously when designing sounds, since it shows the development of the sound with time.
The Low Frequency Oscillator in the K5 has only a few parameters, so you would not expect to find anything unusual here — but there is! Try setting up the LFO as follows:
SHAPE = 3
SPEED = 99
DELAY = 10
TREND = 31
Set the LFO parameter on the DFG page to 31. Now press and hold down a few keys. You will hear what sounds like a manic Morse Code tapper! This sharp burst of harmonics can be very useful, especially if you use a reverb or delay processor on the output of the K5. This almost makes up for the fact that no random/sample-hold waveform is provided. The K5 also has an interesting reset mechanism for the delay — you have to lift all your fingers from the keys (and the K5 then sends an All Notes Off MIDI message!) before it resets the delay. This means that by choosing between legato and staccato, you can control delay effects from the keyboard.
The DFT (it should really be OFF!) is more useful than you might think. Subtle or extreme EQ can make all the difference to a sound. Here are some techniques to try:
Notches — try using a flat (all 63) DFT curve, except for one midrange value which is set to zero. Many real instruments have exactly this type of deep notch structure to their harmonic spectrum, so this technique can be very effective for simulating guitar and bass sounds, but it can also thin out synthetic sounds to give more unusual textures.
Combs — set up alternate bands set to 63, with the remainder set to 0. The resulting series of notches is called a comb filter, and is a more extreme version of the notch described above. Again, this is similar to the responses which some real instruments have, but it is also the basis for the 'phasing' effect often overused on guitars in the Seventies — except that this phasing is fixed instead of cyclically changing in frequency. Comb filters tend to lower the output level of the sound quite drastically, so use them sparingly.
High Pass — many current synths are bringing back low pass filters (the D50, VFX, and M1 all use low pass filters as their main filtering element). Although this gives some nice dynamic sounds, it also brings back a problem which used to occur when people recorded using only analogue synths — the bass end of your sound begins to get over-used. All the synths are filtering just higher frequencies and you get almost no filtering of low notes, resulting in a wooly, muddy sound. By using the DFT to remove all the low frequencies, you can obtain a clear, light sound which does not add to the confusion and can be a refreshingly different timbre to use.
Fundamental Removal — if you filter too much from the mid-frequency ranges you can remove the fundamental and major overtones of the sound you are playing — this has a characteristic effect of suddenly making the sound appear as if it is in a tube, or has moved into the distance. This can be useful if you want to create unusual textural sounds, but can destroy some simulations of real sounds.
Band Pass — this may appear obvious, but try restricting the bandwidth of sounds by removing some top and some bass. One of the interesting things about FM sounds is their unpredictable bandwidths, and you can extend your sound palette quite considerably if some of your sounds have limited frequency responses. Many real instruments have a similar response, especially wind and brass instruments.
The final point to remember about the K5 is that it often helps to make your sounds too bright and then tone down the high frequencies with EQ on your mixer. This helps keep the noise level down — especially useful on an 8-bit machine like the K5. If you are wondering why an 8-bit machine should be interesting in these days of 16-bit and 22-bit megasynths, then you should remember that it is often the dirt and noise which give character to instruments, and the K5 certainly has plenty of personality!
Before I forget, if you have noticed that your K5 seems to lose its system settings (Local On/Off, MIDI Transmit and Receive Channels etc), then do not worry — all K5s do this!
Additive synthesis (as used by the K5) seems to be a programmer's tool rather than a major league bestseller. Unlike the DX7, which sounded so completely different from anything else at the time that everyone had to have one, the K5 suffers from exactly the opposite problem — it can sound like almost anything: pseudo-sampled pianos, brash filter swept analogues, atmospheric evolving textures, and breathy chiffs are just different aspects of its repertoire. This makes it difficult to conveniently label the K5, since it can sound like analogue and digital, FM and LA, sampled and synthesized. As far as I can see, this is not a problem, it is a positive advantage!
I hope that this article encourages you to delve deeper into your K5, or at least to investigate some of the alternative sounds which are available. In these days of instant sounds for popular instruments, using a synth like the K5 can really give you a distinctive and memorable sound — and if you think about it, that's what people remember about Motown, The Beach Boys, The Beatles, Stock-Aitken-Waterman, etc — an individual, instantly recognisable sound. The Kawai K5 may be an unexploited backwater of your synthesis armoury at the moment, but if I were you, I would move it into the front line right now!
K5 Utility program £7.
SOS Software, (Contact Details).
Gear in this article:
Feature by Martin Russ
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!