Title :
An ANN for recognizing melody preferences
Author :
Eaton, Shelby L. ; Stiber, Michael
Author_Institution :
Comput. & Software Syst., Washington Univ., Bothell, WA, USA
Abstract :
This paper summarizes the development of a fully interconnected backpropagation neural network that analyzes users´ musical tastes. This preliminary investigation is intended to discover if this type of network is appropriate for creating a program that will predict which selections a user will enjoy, based on preference observations. Musical selections of 32 notes were chosen and used as inputs to the network. Preferences by tonality (major minor), as well as preferences by musical style (baroque, romantic) were tested. In preference by tonality, the network correctly predicted a user´s preference for major music 66% of the time. In the preference by style, the network was not successful predicting correctly only 33% of the time
Keywords :
backpropagation; music; neural nets; pattern recognition; backpropagation neural network; learning; melody preference recognition; pattern recognition; tonality; user musical tastes; Artificial neural networks; Backpropagation; Cognition; Computer networks; Instruments; Music; Neural networks; Software systems; Testing; Timbre;
Conference_Titel :
Neural Networks, 1999. IJCNN '99. International Joint Conference on
Conference_Location :
Washington, DC
Print_ISBN :
0-7803-5529-6
DOI :
10.1109/IJCNN.1999.836240
