Identifying Metric Types with Optimized DFT and Autocorrelation Models
class CoAuthor
def __init__(self, name):
self.name = "Jason Yust"
Format (When): Where
TL;DR
This paper explores the classification of metric types using different feature representations. Using weighted timepoint, DFT, and autocorrelation, we train feedforward neural networks to distinguish allemandes, courantes, sarabandes, and gavottes in the Yale-Classical Archives Corpus. Autocorrelation and DFT models perform better than a baseline, with DFT consistently better by a small amount.