ESTIMATING THE TONALITY OF POLYPHONIC AUDIO FILES:
COGNITIVE VERSUS MACHINE LEARNING MODELLING STRATEGIES

Emilia Gómez, Perfecto Herrera
Music Technology Group, Institut Universitari de l’Audiovisual, Universitat Pompeu Fabra
{emilia.gomez,perfecto.herrera} at iua.upf.es

Additional material to the ISMIR paper. Here you can find the poster presented during the conference.

Confusion Matrix for the test database using a cognitive model (up) and a machine learning approach (down), for simultaneous key note and mode estimation. The machine learning technique that is used is a multilayer back-propagated perceptron with a single hidden layer of 20 units.

Rows represent the correct key and columns the estimated one. The number of confusions is written in the corresponding cell.

Matrix Cognitive
Matrix Machine Learning
Labels

Back to homepage