Eigentriads and Eigenprogressions on the Tonnetz
This work addresses the problem of improving pattern recognition in music information retrieval for researchers and practitioners, but it is incremental as it builds on existing methods like spiral scattering transform and spectral graph theory.
The paper tackles the problem of characterizing Western tonal harmony patterns in music information retrieval by introducing the eigenprogression transform, a new multidimensional representation that is equivariant to time shifts and pitch transpositions, and reports state-of-the-art results on a supervised composer recognition task (Haydn vs. Mozart) from polyphonic MIDI pieces.
We introduce a new multidimensional representation, named eigenprogression transform, that characterizes some essential patterns of Western tonal harmony while being equivariant to time shifts and pitch transpositions. This representation is deep, multiscale, and convolutional in the piano-roll domain, yet incurs no prior training, and is thus suited to both supervised and unsupervised MIR tasks. The eigenprogression transform combines ideas from the spiral scattering transform, spectral graph theory, and wavelet shrinkage denoising. We report state-of-the-art results on a task of supervised composer recognition (Haydn vs. Mozart) from polyphonic music pieces in MIDI format.