where H is the maximum value of all H i and the indicator function χ hf 0 ( t ) < f s / 2 ( t ) discards any harmonic segment whose frequency is higher than Nyquist.
In contrast, temporal features (for example, duration) of the tonal harmonic segment of the call seem to encode the behavioral vocal cues.
Thereby, the harmonic segment was either not present or of a very short duration in the digging context (mean ± sd; 0.01 ± 0.02 s), while its duration increased in the searching context (0.05 ± 0.03 s).
A group-specific vocal cue was found in the noisy segment of the call (number of correctly cross-classified elements (ncce) = 44.47, P = 0.038, n = 36), but not for the whole call (ncce = 38.08, P = 0.27), nor for the harmonic segment (ncce = 44.47, P = 0.038, n = 36).
The percentage of correct classification after cross-validation (CV) to individuals within each of the four study groups compared to that expected by chance; Results for the whole call, noisy segment and harmonic segment are given; p-values are derived from bootstrapping method [ 46]; p ≤ 0.1, * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001 a Number of individuals tested b Expected by chance.
Specifically, the initial noisy segment of the call remained stable within an individual in all of the quantified behavioral contexts, while a gradation was detected in the subsequent harmonic tonal segment.
Different from the Bayesian NMF (BNMF) [15], BGS-NMF conducts group sparse learning which does not only characterize the within-segment harmonic information but also represent the across-segment rhythmic regularity.
The first group was the common bases which were used to explore the inter-segment repetitive characteristics, while the second was the individual bases which were applied to represent the intra-segment harmonic information.
For units in segment, the harmonic orders are in accordance with segment harmonic order (14).
