Homepage of Arno van Goch


AvSound is an old experimental DOS-program that was originally written for the Ensoniq EPS 16+ Digital Sampling Workstation, but it can be of use to anyone who likes experimenting with samples and sounds.

AvSound takes a wav-file, and converts it to another wav-file that represents something that could best be described as 'the average sound' of the original soundfile. The program is freeware and can be downloaded here

How it works

AvSound does not use spectrum analysis, but uses a brute force method. The procedure is as follows : The destination soundfile (usually longer than the source) is treated as an looped tape. The source is mixed in at random starting points. After say 1000 mixes the individual mixes cannot be distinguished and the process is finished.

This procedure introduces the following features :

  • The resulting soundfile has a perfect loop
  • The resulting soundfile can be of any size, although longer loops work best, because this way you won't recognize a repetition in the sound
  • Because of the randomness of the mixing, the resulting soundfile sounds monotonous yet alive

As you might understand, all information about attack and decay is lost so, if you want to, you will have to try recreating it by using envelopes. Also the final result largely depends on the source. A single tone which fades in and out works best.

Demo sounds

Below are some demo sounds. In each example you first hear the sample that is used as an input for AvSound. After that you will hear the resulting output of AvSound, followed by that same output but this time played as a loop.


To get things really interesting I use the following technique: I take one source wavesample and generate two destination wavesamples. One of 100000 samples and one of say 100093 samples. I pan the first wavesample a little to the left, and pan the second a little to the right. Also I detune the wavesamples a little. Because the wavesamples are not exactly the same, the detuning doesn't create a synthetic sounding phaser effect. And because the two wavesamples run with two different loops, it takes some time for the whole sound to repeat in exactly the same way, so what I get is a ever changing sound.

After working with the program for a while, I found that it's best use is to create soft background sounds. Usually I apply a lowpass- filter to the resulting wavesample.


To use the program you will need to have some knowledge of using the windows command prompt. Also note that the program has some serious sample-size limitations according to today's standards.