Detection Possibilities of Depression and Parkinson’s disease Based on the Ratio of Transient Parts of the Speech
Gábor József Kiss, Artur Bendeguz Takacs, Dávid Sztahó, Klára Vicsi · 2018
Speech production is a complex and complicated brain process. Thus the speech of subjects who suffer from neurological or psychological disease such as Parkinson's disease or depression differ from the normal speech from healthy speakers. It is characteristic, that their speech tempo is slower, and their articulation is inadequate. These alterations modify the speech product to become monotone and reduce the rate of rapid and significant changes. In this study it was investigated how the ratio of transient parts is affected in case of depression and Parkinson's disease. Overall 321 speech samples were examined from healthy, depressed and Parkinson's diseased speakers. The transient parts were detected in their speech using an automatic transient detection method, and based on the detection, the ratio of the transient parts (RoT) were calculated. It was found that the mean value of RoT in case of depressed and Parkinson's diseased speakers decreased by 9-10% compared to healthy speakers. The RoT was used as the only input of classification experiments, and it achieved quite promising results for separation of healthy and non-healthy speakers with accuracy of 81%.