A Real-Time Timbre Tracking Model Based on Similarity

for a masters degree of the Institute of Sonology, The Hague. With the introduction of the computer as a musical tool, the role of tim-bre in composition has been largely emphasized. Until today, many sound synthesis models have been introduced, each one offering a different repre-sentation for timbre. Nevertheless, current definition of timbre reveals our relatively poor understanding about what timbre is. When we try to extract timbral properties from an acoustic sound, this becomes crucial. The real-time timbre tracking model presented in this thesis tackles this difficult issue. By following the essence of Timbre Space technique exten-sively used in the non real-time timbre recognition field, the model tracks timbre of a monophonic acoustic instrument. Unlike conventional approaches, in this model, timbre is described by its similarity/dissimilarity. The model is fully implemented in the Max/MSP programming environ-ment, and can be applied for digital instrument designs and in interactiv