Comparative analysis of multiple speech tasks to recognise Parkinson's disease using pre-trained feature extractor embeddings
Parkinson’s disease is one of the most common neurological diseases, which is currently incurable. Speech can be a suitable biomarker for supporting the diagnosis of the disease. Therefore, many speech tasks and recording lengths are used widely in the literature. Our research compares the recogniti...
Elmentve itt :
| Szerzők: | |
|---|---|
| Testületi szerző: | |
| Dokumentumtípus: | Könyv része |
| Megjelent: |
Szegedi Tudományegyetem TTIK, Informatikai Intézet
Szeged
2024
|
| Sorozat: | Magyar Számítógépes Nyelvészeti Konferencia
20 |
| Kulcsszavak: | Parkinson-kór, Gépi tanulás, Nyelvészet - számítógép alkalmazása |
| Tárgyszavak: | |
| Online Access: | http://acta.bibl.u-szeged.hu/88761 |
| Tartalmi kivonat: | Parkinson’s disease is one of the most common neurological diseases, which is currently incurable. Speech can be a suitable biomarker for supporting the diagnosis of the disease. Therefore, many speech tasks and recording lengths are used widely in the literature. Our research compares the recognition performance measured on seven speech tasks using pre-trained out-of-domain feature extraction algorithms (x-vector, e-capa). We also examine how the voting on the speech tasks relates to the performance on the given speech tasks and which speech tasks the classifier considers essential. Our results showed that using a longer speech signal provides better recognition, and the type of the task is essential, e.g. pronouncing syllables. Furthermore, the classifier considers longer speech tasks more critical in the decision and suggests letting out sustained vowels. |
|---|---|
| Terjedelem/Fizikai jellemzők: | 173-186 |
| ISBN: | 978-963-306-973-8 |