Table 2. Description of the “Mobile Device Voice Recordings at King’s College London (MDVR-KCL)”.
The number of speakers is 37, split between healthy and sick patients. Furthermore, the gender and the UPDRS II-5 score are introduced in the Table. It is possible to observe how unbalanced the dataset is, particularly regarding gender and the UPDRS II-5 score. For each speaker, the dataset provides two sets of recordings. In our experiments, we use the read text and set the scenario to a text-dependent one. Moreover, we conduct our analysis by patient, and therefore we will be in a speaker-dependent setting.
| MDVR-KCL Dataset Description | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Parkinson’s disease Status | Healthy | Sick | ||||||||
| Gender | Female | Male | Female | Male | ||||||
| UPDRS II-5 score | − | − | 0 | 1 | 2 | 3 | 0 | 1 | 2 | 3 |
| # of Speakers | 19 | 2 | 2 | 2 | − | − | 5 | 4 | 2 | 1 |