Modellvorstellungen über Klangfarbe: Von der »Manichfaltigkeit der Praedicate« zum Timbre Space

What are the timbral characteristics of musical sounds? From the second half of the nineteenth century, ideas about musical timbre were affected by contradictory and platitude statements that have lead researchers and musicians throughout history to adopt various models and perspectives. However, the influence of pitch and dynamics has largely been ignored within these models by definition. Another fact had further effects on the history of psychoacoustical research: While Hermann von Helmholtz’s Lehre von den Tonempfindungen was soon translated into English by Alexander J. Ellis, a translation of Carl Stumpf’s standard works Tonpsychologie (1883 and 1890) and Die Sprachlaute (1926) has yet to be published, as are translations of the important writings of other German-speaking successors to Helmholtz and Stumpf (such as Köhler, Schumann, Albersheim, Reinecke, Fricke etc.). Consequently, Anglo-American psychoacoustical research, which has been most influential in this field since World War II, is lacking this particular aspect of knowledge about musical timbre. The relevance of the German research tradition comes to light when we examine an instrument in its entire pitch and dynamic range (rather than only one or a few pitches as in most timbre studies): timbre perception of musical instruments is determined to a certain extent by steady pitch-independent formant areas, whose origin and characteristics at pitch and dynamic changes has generally been resolved today. These timbral effects appear to be in accordance with the principles of Auditory Scene Analysis (Albert S. Bregman) as well as with recommendations in orchestration treatises of the last centuries. Anglo-American models, such as John Grey’s Timbre Space and subsequent research, and German models of timbre perception both have advantages and short- coming. With the help of sound analysis and re-synthesis software and in combination with the concept of mel-frequency cepstral coefficients methods, it seems possible to merge these diverse approaches into a common timbre perception model