In this paper we try to identify spectral and acoustic features that are distinctive of Parkinson's disease patients' speech. We investigate the contribution of several features' families to a simple classification task that distinguishes between two balanced groups-patients with Parkinson's disease and their age and gender matched group of Healthy Controls, both uttering sustained vowels. We achieve over 75% correct classification using a combination of acoustic and spectral features. We show that combining a few statistical functionals of these features yields very good results.. This can be explained by two reasons: the first is that the statistics of Parkinson's disease patients' speech defer from those of Healthy people's speech; the second and more important one is the gradual nature of the Parkinsonian speech that is manifested by the changes within an utterance. We speculate that the feature families that most contribute to the classification task are the most distinctive for detecting the disease and suggest testing this hypothesis by performing long-term analysis of both patient and healthy control subjects. Similar accuracy is obtained when analyzing spontaneous speech where each utterance is represented by a single normalized i-vector.

This paper deals with clustering of speakers' short segments, in a scenario where additional segments continue to arrive and should be constantly clustered together with previous segments that were already clustered. In realistic applications, it is not possible to cluster all segments every time a new segment arrives. Hence, incremental clustering is applied in an on-line mode. New segments can either belong to existing speakers, therefore, have to be assigned to one of the existing clusters, or they could belong to new speakers and thus new clusters should be formed. In this work we show that if there are enough segments per speaker in the off-line initial clustering process, it constitutes a good starting point for the incremental on-line clustering. In this case, incremental online clustering can be successfully applied based on the previously proposed mean-shift clustering algorithm with PLDA score as a similarity measure and with k-nearest neighbors (kNN) neighborhood selection.

No abstract

No abstract

No abstract