TABLE 2.
Categorization of the 65 Low-Level Descriptors (LLDs) of the ComParE Feature Set and Specification of Involvement (√) or Non-Involvement (×) in a Top Feature of the Respective Differentiation Task (Task 1: pos vs neg, Task 2: pos+ vs neg+, and Task 3: pos– vs neg–).
| Group | Energy-related LLDs (4) | Task 1: pos vs neg |
Task 2: pos+ vs neg+ |
Task 3: pos– vs neg– |
|---|---|---|---|---|
| Prosodic | Auditory spectrum sum (loudness) | √ | √ | √ |
| Prosodic | RASTA-filtered auditory spectrum sum | √ | √ | × |
| Prosodic | RMS energy | √ | √ | √ |
| Prosodic | zero-crossing rate | × | √ | × |
| Group | Spectral LLDs (55) | pos vs neg | pos+ vs neg+ | pos– vs neg– |
|---|---|---|---|---|
| Cepstral | MFCC 1–14 | √ | √ | √ |
| Spectral | Psychoacoustic harmonicity | √ | √ | √ |
| Spectral | Psychoacoustic sharpness | × | √ | × |
| Spectral | Spectral centroid | × | √ | × |
| Spectral | Spectral energy 250–650 Hz, 1–4 kHz | √ | √ | √ |
| Spectral | Spectral entropy | × | √ | × |
| Spectral | Spectral flux | √ | √ | √ |
| Spectral | Spectral kurtosis | √ | √ | × |
| Spectral | Spectral roll-off point 0.25, 0.50, 0.75, 0.90 | √ | √ | × |
| Spectral | Spectral skewness | × | √ | × |
| Spectral | Spectral slope | √ | √ | √ |
| Spectral | Spectral variance | × | √ | × |
| Spectral | RASTA-filtered auditory spectral band 1–26 | √ | √ | √ |
| Group | Voicing-related LLDs (6) | pos vs neg | pos+ vs neg+ | pos– vs neg– |
|---|---|---|---|---|
| Prosodic | Fundamental frequency | × | √ | × |
| Quality | HNR | √ | √ | √ |
| Quality | Jitter (local and DDP) | × | √ | × |
| Quality | Shimmer | × | √ | × |
| Quality | Voicing probability | × | √ | × |
DDP, difference of differences of periods; HNR, harmonics-to-noise ratio; MFCC, Mel-frequency cepstral coefficient; neg, COVID-19 negative; pos, COVID-19 positive; RASTA, relative spectral transform; RMS, root mean square; +, symptomatic; –, asymptomatic.