Skip to main content

Table 8 Generalization—Combine the acoustic features using PCA

From: Performance of machine learning algorithms for dementia assessment: impacts of language tasks, recording media, and modalities

Feature Name

Functional

Principle Component (PC)

MFCC 0-14

Mean

1st PC from the means of 15 MFCCs

  

2nd PC from the means of 15 MFCCs

 

Kurt

1st PC from the kurt of 15 MFCCs

  

2nd PC from the kurt of 15 MFCCs

 

Skew

1st PC from the skew of 15 MFCCs

  

2nd PC from the skew of 15 MFCCs

ΔMFCC 0-14

mean

1st PC from the means of 15 ΔMFCCs

  

2nd PC from the means of 15 ΔMFCCs

 

Kurt

1st PC from the kurt of 15 ΔMFCCs

  

2nd PC from the kurt of 15 ΔMFCCs

 

Skew

1st PC from the skew of 15 ΔMFCCs

  

2nd PC from the skew of 15 ΔMFCCs

ΔLSP freq 0-7

mean

1st PC from the means of 8 ΔLSP freq

  

2nd PC from the means of 8 ΔLSP freq

 

Kurt

1st PC from the kurt of 8 ΔLSP freq

  

2nd PC from the kurt of 8 ΔLSP freq

 

Skew

1st PC from the skew of 8 ΔLSP freq

  

2nd PC from the skew of 8 ΔLSP freq