Modulation spectra for automatic detection of Parkinson's disease

In this paper, we explore the information provided by a joint acoustic and modulation frequency representation, referred as modulation spectrum, for detection of people with Parkinsons disease through speech signals. The set of features includes the centroids and the energy content of different frequency bands in the modulation spectra of the recordings. Additionally, with the aim to eliminate possible redundancy in the information provided by the features, two different feature extraction techniques are applied, Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA). The classification was done by means of Gaussian mixture model (GMM). The results show that this approach is acceptable for this purpose, with the best accuracy around 71% for vowel /i/.