Table 4.
Features with the highest and lowest ICC per category
| Category | Highest ICC | Lowest ICC |
|---|---|---|
| Energy | pcm_LOGenergy_sma_de_de_linregerrA (ICC = 0.93) pcm_LOGenergy_sma_de_de_nzabsmean (ICC = 0.93) pcm_LOGenergy_sma_meanPeakDist (ICC = 0.94) |
pcm_LOGenergy_sma_de_centroid (ICC = 0.39) pcm_LOGenergy_sma_de_de_qregerrA (ICC = 0.39) pcm_LOGenergy_sma_numPeaks (ICC = 0.23) |
| Frequency | F0semitoneFrom27.5Hz_sma3nz_amean (ICC = 0.89) F0semitoneFrom27.5Hz_sma3nz_percentile50.0 (ICC = 0.92) F0semitoneFrom27.5Hz_sma3nz_percentile80.0 (ICC = 0.88) |
F0semitoneFrom27.5Hz_sma3nz_meanFallingSlope (ICC = 0.28) F0semitoneFrom27.5Hz_sma3nz_stddevFallingSlope (ICC = 0.21) F0semitoneFrom27.5Hz_sma3nz_stddevRisingSlope (ICC = 0.24) |
| MFCC | mfcc_sma_de_de.0._percentile95.0 (ICC = 0.95) mfcc_sma.12._meanPeakDist (ICC = 0.98) mfcc_sma.12._qregerrA (ICC = 0.94) |
mfcc_sma_de_de.11._centroid (ICC = 0.20) mfcc_sma_de_de.11._linregc1 (ICC = 0.18) mfcc_sina_de_de.8._quarrile2 (ICC = 0.06) |
| Pitch | voiceProb_sma_linregerrQ (ICC = 0.89) voiceProb_sma_qregerrA (ICC = 0.90) voiceProb_sma_qregerrQ (ICC = 0.91) |
F0_sma_quartile1 (ICC = 0.09) F0env_sma_de_de_quartile2 (ICC = 0.02) F0env_sma_de_quartile2 (ICC = 0.00) |
| Spectral | pcm_fftMag_melspec_sma.1._meanPeakDist (ICC = 0.92) pcm_fftMag_melspec_sma.2._meanPeakDist (ICC = 0.92) pcm_fftMag_melspec_sma.2._qregerrA (ICC = 0.93) |
mfcc2_sma3_stddevNorm (ICC = 0.00) mfcc4_sma3_stddevNorm (ICC = 0.00) slopeV0.500_sma3nz_stddevNorm (ICC = 0.00) |
| Temporal | loudnessPeaksPerSec (ICC = 0.94) pcm_zcr_sma_de_de_numPeaks (ICC = 0.92) pcm_zcr_sma_de_numPeaks (ICC = 0.92) |
pcm_zcr sma_de_de_amean (ICC = 0.27) pcm_zcr sma_de_de_qregc1 (ICC = 0.21) pcm_zcr sma_de_de_qregc2 (ICC = 0.25) |
| Lexical | brunet (ICC = 0.71) MATTR_50 (ICC = 0.59) TTR (ICC = 0.75) |
adverbs (ICC = 0.00) coordinate (ICC = 0.00) familiarity (ICC = 0.00) |
| Pragmatic | rst_num_attribution (ICC = 0.39) rst_num_background (ICC = 0.39) rst_num_cause (ICC = 0.39) |
topic18 (ICC = 0.00) topic20 (ICC = 0.00) topic22 (ICC = 0.00) |
| Semantic | avg_wn_ambig (ICC = 0.69) kurt_wn_ambig_vb (ICC = 0.72) sd_wn_ambig (ICC = 0.64) |
avg_max_wn_depth_vb (ICC = 0.00) avg_min_wn_depth_vb (ICC = 0.00) avg_wn_sim_LC (ICC = 0.00) |
| Syntactic | S (ICC = 0.73) VP_.._VB_VP (ICC = 0.50) W (ICC = 0.67) |
ADJP_.._JJ (ICC = 0.00) ADJP_.._RB_JJ (ICC = 0.00) ADVP_.._RB (ICC = 0.00) |
For information about the features, readers can refer to openSMILE and Praat for the acoustic features and Talk2me for the language features.