Columns

Beginners

Synthesis

Processing

Real-Time

Features

Woodwinds

Departments

Links

Editorial

Contribute

Back Issues

Editorial

Hans Mikelson

hljmm@werewolf.net

With this issue I have completed the first year of Csound Magazine. Somewhat of a milestone. I have been looking back at the past issues as I worked on this issue. Many people have contributed a great amount of work to make the magazine a success. Every issue contributers have come through with excellent articles. I want to thank the contributers for their support. Sometimes I worry that I am becoming too much of a pest when I solicit people for articles. I will try to avoid this if possible in the future but if you find me bothersome please let me know and I will stop bugging you. I realize I do not offer much in terms of compensation for your hard work except a little web space, my appreciation and the appreciation of the readers.

There was some discussion on the mailing list regarding designing an instrument. This is the type of thing I like talking about so since this is the editorial page I will. I use several different approaches to sound design.

One of my favorites is to find a mathematical equation or concept which has not been applied to sound before (as far as I know) and try it out to see what it sounds like. I generally keep a look out for interesting equations, functions and systems. I never know before hand exactly what it will sound like. Often I end up with surprisingly pleasing results. Even if the results do not sound good at first they can almost always be tweaked into sounding good with the approriate choice of frequency, filters and other parameters.

Another method I like to use is to select natural sounds and try to synthesize them. I did this in my recent orchestra for crickets. In the past I attempted the same for ocean waves, birds, wind and wolf howls. When I am attempting this type of synthesis I often get a sample of the target sound and look at it in terms of the frequency content, the overall envelope, the waveform and other aspects of the sound. I try to see if the frequencies are broadband and therefore noise based or narrow band and therefore more like sine waves. Broadband sounds can be approximated by filtered noise, FM with noise as modulator and formant synthesis. Narrow band sound can be produced with sine or other oscillators. Many natural sounds have some type of amplitude pulsing or different aspects which fade in and out during the sound. I first try to create the sound in any way possible perhaps by using several i-events to make a single sound. Later I try to make the sound more usable by combining what was several instruments or instrument calls into a single instrument. I go through the input parameters and try to make them so they are very usable. This is often different from what is easiest to code.

A third method I use is to try to develop some type of model of physical phenomenon and to implement this with DSP elements. Perhaps the way the sound is produced suggests a type of filter, waveguide or some other system.

A fourth method is to find a sound which is similar to the sound you want and modify it until it sounds the way you want.

Along the way many happy accidents occur. You create a sound that was completely unlike what you were trying for but sounds incredibly cool. This seems to happen quite a bit with Csound. Then there is the unhappy accident. When you have a very nice sound but want to improve some small aspect of it. The more changes you make the further you get away from the original sound until it is lost and you cannot remember how to get it back to sounding good.

Listen, learn and pay attention to sounds around you. Study how the sounds are produced. Use all of the methods above where appropriate. If one approach is not working try a different approach.

This issue I again tackle the columns with articles on panning, drum sounds, reverb and Keykit. Josep Comajuncosas came through with a very nice feature on modeling woodwind instruments that presents some innovative solutions to some of the problems encountered when working with physical models.