Skip to main content

Table 2 Two hundred five acoustic features in the proposed feature set (EmoFt): low-level descriptors (LLD) and functionals (brute-force combination) as well as features derived from the long-term average spectrum (LTAS) and three other features(see text)

From: Emotion in the singing voice—a deeperlook at acoustic features in the light ofautomatic classification

19 LLD

Loudness, spectral flux and entropy

Energy in bands 0–0.5 and 0–1 kHz

Slope of log. power spectrum 0–1, 0–5, and 1–5 kHz

Alpha ratio (in dB), Hammarberg index (in dB)

MFCC 1–4, harmonics-to-noise ratio

F 0, prob. of voicing, jitter, and shimmer (local)

Five functionals

Arithmetic mean, standard deviation

Fifth and 95th percentile and range 5–95 %

Long-term average spectrum (LTAS), 27 bands

MFCC 1–4, spectral entropy

Energy in bands 0–0.5 and 0–1 kHz

Slope of log. band spectrum 0–1, 0–5, and 1–5 kHz

Alpha ratio (in dB), Hammarberg index (in dB)

Others

Equivalent sound level (in dB)

Frequency with maximum amplitude in modulation spectrum of

F 0 and loudness