Research Article

FPT-Former: A Flexible Parallel Transformer of Recognizing Depression by Using Audiovisual Expert-Knowledge-Based Multimodal Measures

Table 2

Selected eGeMAPS features for depression diagnosis.

Feature nameFeature declaration

LoudnessThe overall volume or sound intensity of a sound signal
Alpha ratioThe energy ratio of the sound signal spectrum’s low and high-frequency parts
Hammarberg indexThe change pattern of the fundamental frequency in the sound signal
Slope 0–500The frequency spectrum’s rate of alteration is assessed in the range from 0 Hz to 500 Hz
Slope 500–1500The frequency spectrum’s rate of alteration is assessed in the range from 500 Hz to 1500 Hz
Spectral fluxThe amount of flow or variation in the spectrum of a sound signal
mfcc1The first Mel-Frequency Cepstral Coefficients
mfcc2The second Mel-Frequency Cepstral Coefficients
mfcc3The third Mel-Frequency Cepstral Coefficients
mfcc4The fourth Mel-Frequency Cepstral Coefficients
F0semitoneFrom27.5 HzThe semitone difference between the fundamental frequency of the sound signal and 27.5 Hz
Jitter localThe local jitter of the sound signal
Shimmer local dBThe local trill of a sound signal
HNRdBACFHarmonic to noise ratio (HNR) of a sound signal
LogRelF0-H1-H2The logarithmic difference between the fundamental frequency in the sound signal and the corresponding first (H1) and second (H2) harmonics
logRelF0-H1-A3The logarithmic difference between the fundamental frequency in the sound signal and the corresponding first harmonic (H1) and third formant (A3)
F1frequencyThe frequency of the first formant (F1) in the sound signal
F1bandwidthThe bandwidth of the first formant (F1) in the sound signal
F1amplitudeLogRelF0The logarithmic difference between the amplitude of the first formant (F1) in the sound signal and the fundamental frequency
F2frequencyThe frequency of the second formant (F2) in the sound signal
F2amplitudeLogRelF0The logarithmic difference between the amplitude of the second formant (F2) in the sound signal and the fundamental frequency
F3frequencyThe frequency of the third formant (F3) in the sound signal
F3amplitudeLogRelF0The logarithmic difference between the amplitude of the third formant (F3) in the sound signal and the fundamental frequency