Uebertragungsfehler

Diemo Schwarz
schwarz@ircam.fr

April 29, 1998



Artistic Concept

Übertragungsfehler (transmission error) starts out as a reflection on the various types of errors that can occur in the process of creating music by means of digital sound processing with computers. Right at the beginning of the piece stands a software error: a sound editing program failing to play back what it is supposed to play, instead playing the contents of the computer's memory.

Most of the following sounds have been filtered or processed through images, by an algorithm I wrote, which uses the intensity levels of the pixels to obtain the spectrum of a filter which was applied to the sound source. This procedure bears a severe methodological error, however, since an image has no meaning in musical or signal processing terms.

The center and pivoting point of the piece is an extended fractal structure, inspired by the fact that the mathematician Benoit Mandelbrot developed the very notion of fractals after studying the behaviour of error patterns in communication lines. Thus, transmission errors are at the heart of the theory of fractals, deterministic chaos, and the beautiful structures of the Mandelbrot or Julia set. Yet, the fractal structure used in this piece (figure 1) is erroneous.



Figure 1: Fractal structure in Übertragungsfehler.



Figure 2: Waveform resulting from the fractal structure.

The question asked in Übertragungsfehler is: How can we distinguish the inherent errors in the digital process (quantisation noise, aliasing) from technical errors (drop outs, software errors) from operating errors (clicks, clipping, distortion) from methodological errors --- from intended errors. Isn't the error at the heart of the creative process itself?1

Even everyday communication is riddled with transmission errors, as Umberto Eco has shown, who pointed out that transmitter and receiver never operate with a congruent set of symbols. Nevertheless, communication is successful.

Transmission errors are thus a major driving force behind all development, and the conscious treatment of such errors should be a prominent concern for the artist. Instead of eliminating such errors, my objective has therefore been to highlight and exploit them.

Technical Notes

Various programs and sound sources were used to create Übertragungsfehler. My research project at IRCAM, spectral envelope analysis of sound and application to additive synthesis, yielded as an offshoot a program to convert a spectral envelope obtained by the discrete cepstrum method into an image and back. The spectral envelope of a mongolian chant (see figure 3) is thus fed as an image into Metasynth, where it is interpreted completely differently---another methodological error of incompatibility of models---but yielding an overwhelming sonic result.



Figure 3: Spectral Envelope of a mongolian chant.

I also wrote a program to convert an image into a description of filter transfer functions over time, usable by IRCAM's phase vocoder SVP. This was used to filter an old recording not created specifically for this composition, which I therefore consider a found object, with the image in figure 5, which was created by manipulating the original image (see figure 4) with Photoshop.



Figure 4: The original picture.



Figure 5: The picture used for filtering.

The fractal structure in figure 1 was created by hand in Cubase VST, assigning a specific stereo position and EQ setting to each track.

1
In the biggest and most important creative process on earth, the evolution of life itself, transmission errors are beyond doubt the very basis of progress.
This document was translated from LATEX by HEVEA.