Supplementary information data The dataset includes the locations of the 997 language varieties along with their corresponding humidity values (g/kg), jitter rates, shimmer rates, number of tones, language family membership, and raw acoustic data (jitter and shimmer for each audio file). Each language variety is represented by about 1200 sound files, amounting to a total of 1,174,686 recordings. Some lexical items have two or more expressions, all recorded in a single recording. We segmented different expressions. Thus, a recording can yield more than one jitter or shimmer datum.