Skip to main content

Advertisement

Table 7 Acoustic features used in the A-Hybrid DTW+LVCSR QbE STD system

From: Search on speech from spoken queries: the Multi-domain International ALBAYZIN 2018 Query-by-Example Spoken Term Detection Evaluation

Description Number of features
Sum of auditory spectra 1
Zero-crossing rate 1
Sum of RASTA style filtering auditory spectra 1
Frame intensity 1
Frame loudness 1
Root mean square energy and log-energy 2
Energy in frequency bands 250–650 Hz (energy 250–650) and 1000–4000 Hz 2
Spectral Rolloff points at 25%, 50%, 75%, 90% 4
Spectral flux 1
Spectral entropy 1
Spectral variance 1
Spectral skewness 1
Spectral kurtosis 1
Psychoacoustical sharpness 1
Spectral harmonicity 1
Spectral flatness 1
Mel-frequency cepstral coefficients 16
MFCC filterbank 26
Line spectral pairs 8
Cepstral perceptual linear predictive coefficients 9
RASTA PLP coefficients 9
Fundamental frequency (F0) 1
Probability of voicing 1
Jitter 2
Shimmer 1
log harmonics-to-noise ratio (logHNR) 1
LCP formant frequencies and bandwidths 6
Formant frame intensity 1
Deltas 102
Total 204
  1. PLP perceptual linear predictive, LPC linear predictive coding