Acoustic Modelling of musical instruments...

Using a modular analogue synthesiser  

This part of the site describes a series of interesting simple experiments I did, in the early eighties, aimed at modelling the acoustics of a variety of musical instruments electronically using an extended modular analogue synthesiser. The audio results are presented in the form of short sound clips generated by the patches described in this section. See the “synthesiser adventures” section for pictures and details of the the synthesiser I built to do it.

By doing this I hoped to generate sounds that sounded like REAL instruments, rather than electronic sounding ones and then just play around with these models to see what could be done. At that time the modelling of the acoustics of real instruments was only just beginning. It was being done by a few researchers in big institutions using software running on huge mainframe computers.  Nice, if you had access to these expensive esoteric machines but I didn’t. However, I reckoned, physics is physics no matter how you do it so I thought I could do it in the analogue world by designing of some special analogue delay line modules to add to my existing instrument. Crazy idea?

Well judge for yourself. The results created a huge amount of interest at the time leading to articles in national magazines, talks and presentations. In this section is a web friendly version of a presentation/demonstration on it that I gave at the International Computer Music Conference at Cologne in 1988.
[If you would like to read the original transcript, here’s an Adobe pdf version...  ICMC1988.PDF ]. 204KB
Intro...
ACOUSTIC MODELLING SOUND EXAMPLES...
The audio results, of course, are the proof it all worked and they are included progressively throughout this section in the form of several short streaming audio MP3 (128kb/sec encoded) examples from the above ICMC presentation. I think you will be quite surprised  by just what can be done with a old home made analogue modular synthesiser.
DIGITAL RE-CREATIONS DOWNLOADS ...
As it was around thirty years ago when I did these experiments, I thought it would be a good idea to ‘celebrate’ if that’s the right word, by attempting to re-create digital versions of some of these basic patches as zip files for free download. If  you have either “Tassman” 4 or Native Instrument’s “Reaktor”, you can now try them out for yourself. For historical reasons, I’m keeping them simple and as close to the originals as possible. I have to say I struggled with “Tassman”. Although it’s a fine application (sorry app), it’s direction is quite different than mine was at the time, but I did get most of the the basic sounds to work in a fashion but on a very limited range of notes. Fortunately I fared much better with Reaktor”. Its basic ‘nuts and bolts’ approach gave me the opportunity to construct basic macros very close to some of my modules particularly the key delay line described below. These patches mimic my original models quite closely and super-cleanly. If you’re able to, I hope you enjoy trying them out!  It’s unfortunate that Native Instruments only allow users patches to be played within the “Reaktor” shell. If I could have done them as VST plugins, I would.
TASSMAN 4 patches and *.txf files; All zipped files here in one 48KB download.

NATIVE INSTRUMENTS REAKTOR 4 patches *.ens files; Zipped download  files  in groups:-
	Strings zip (36KB),   Gongs zip (61KB),   Brass zip (17KB),  Wind zip (36KB).,
SPECIAL DELAY LINE MODULE...
Above is a simple block diagram representation of the modules that were the key additions to my existing analogue modular synthesiser.

The module is a voltage controlled  (Hz/volt) analogue delay line with an overload signal limiter an output low-pass filter and a feedback loop. Its delay time varies over about 3 octaves in the millisecond region, meaning that it can resonate at fundamental frequencies in the region of a few hundred hertz. As this is a sampling device on the input and output, not shown, are anti-aliasing filters required to prevent high frequency signal modulation effects with the sampling frequency. The chip at the centre of it works like this.... The input signal is sampled at high frequency, much like digital audio, but with a voltage controlled frequency (about 25kHz- 200kHz),and the samples are analogue. They are stored in and stepped through a series of analogue capacitive storage elements contained in the chip until they emerge at the other end. The operation is called a bucket brigade or charge coupled device (CCD).  I used 2 devices; the TDA1022 with 512 steps and the TDA1097 with 1536 steps ,which was ideal for bass instruments (neither of these devices are made today).So the quicker samples get stepped through the shorter the delay. The high frequency voltage controlled oscillator generates the stepping commands to the chip. It has to have a very accurate linear frequency/voltage relationship to be remain in tune throughout its range. My whole modular synthesiser’s voltage control system is linear (Hz/volt), Don Buchla’s big modular synthesisers used this by the way. Remember him?

To this basic delay function I included a two input mixer on the front, a built in soft limiting diode based device (to gently limit the signal amplitude before it clips) and a variable first order low pass filter (6dB/oct) with manual control. On the output (not shown) is an optional output signal inverter switch, which is given special mention in the text. Feedback amount has a very fine manual control particularly nearing the 100% feedback point. The more feedback the more the device will ring. Above 100% the device will take off and oscillate so  having a very fine control close to the 100% point is very important.
THE CVP MODULE...

The other slightly unusual module I have is a CVP. It stands for “Control Voltage Processor”. It’s really a modified standard VCA (voltage controlled amplifier). However in the CVP, with zero control voltage on the control input, the gain is unity, so it goes through unaltered. A negative control voltage reduces the gain and a positive control voltage increases the gain. The single control knob adjusts the attenuation on the control signal. A course/fine switch gives the option of modulation degree, which makes controlling frequency modulation, fine vibrato etc. predictable. In the more common logarithmic based synthesisers where control voltages are simply added, multiplying modulation effects are automatic, as adding logs is the same as multiplying. You need a lot less patch cords too. Whereas in a linear Hz/volt system there are no logs so for each  multiplication you need a separate multiplier device. Tuning can be very temperature stable with a linear system as the temperature sensitive circuits used everywhere for the log conversions are absent. This can be very important during concert performance!
|  Home  |  Next page  |
ACOUSTIC ANALOGUE MODELLING: THE MODULES...
You can also compare these with digital versions I created using “Reaktor”, see text opposite. The digitally created sound clips are presented in a new section “Digital Acoustic modelling” at the end of this section. It presents the “Reaktor” panels with equivalent sound clips of the sounds the patches can produce. I hope you find it interesting.
ANALOGUE ...
DIGITAL...