Application of Praat's "To FormantPath..." command for getting by-speaker, by-token formant ceilings, as a more precise alternative to setting general ceilings based on assumptions about speaker sex. Generates numerous formant measurements per token, which may later be aggregated as desired (e.g., by-token medians).

Demoed in get_formant_path.ipynb for single and dual speaker recordings. Requires parselmouth, phonlab, and audiolabel libraries.

Visualization using various by-token grouping stragies and kernel density plots demoed in visualize_formants.ipynb.

Script for dealing with pairwise results for categorial predictors from an R-based linear mixed-effects model for one formant (e.g., F1), a continous outcome: manage_pairwise_results.ipynb

Outlier detection (plotting, saving set of outliers for all speakers): vowel_outliers_05-23-25.ipynb. Contact repo owner for full formants data file.