Exploring the sound structure of novel vocalizations (Repository for paper submitted to Evolang XV)

Folder structure

The repository contains RMarkdown with code to analyse sound similarity in novel vocalizations.

├── data
├── datasets
└── plots

The repository contains:

1. data - file with duration of each segment (the original .wav files are not publicly available)
2. datasets - datasets that have been used to perform analysis and make plots
3. plots - note that the _sns plots (that are also used in the paper) were created using a Python script.

In the datasets , you can find the following files:

• distance_matrix: matrix of distances between a segment and all other segments
• segments_soundgen_summary: ouput of soundgen() function
• umap_df - dataset of concepts and their coordinates in reduced 2-dimensional space
• meanDist_pertrial: complete dataset with all metadata, used features, plus mean calculated from 'distance_matrix' within a trial (i.e., concrete realization of a concept by a participant)
• meanDist_perpcn: complete dataset with all metadata, used features, plus mean calculated from 'distance_matrix' within a participant (ignoring trial differentiation)

Contact

fuchs@leibniz-zas.de / kadava@leibniz-zas.de