Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

< Back to Article

Characterizing Vocal Repertoires—Hard vs. Soft Classification Approaches

Fig 2

Unsupervised Ward’s clustering of 912 chacma baboon calls based on different frequency dependent and temporal feature setups.

The x-axis represents groups of calls, and the y-axis represents average Euclidian within-cluster linkage distance. (A) Set consisting of 118 features. High-frequency (cluster 1) and low-frequency (cluster 2) were segregated into two first-order clusters. High frequency calls further subdivide into more tonal (cluster 1.1) and relatively noisier (cluster 1.2) calls. Low frequency calls subdivide into short and very low-frequency grunt-calls (cluster 2.2), moderate-frequency and harmonic weaning-calls (cluster 2.1.1), and more noisy, short bark-calls (cluster 2.1.2). (B) Set consisting of 38 features. (C) Set consisting of 9 features. (D) Set consisting of 19 factors determined by factor analysis.

Fig 2