Bangla-Align
Bangla-Align is a Python-based toolkit that extends the Montreal Forced Aligner’s capabilities specifically for Bangla speech processing. Designed for speech researchers, phoneticians, phonologists, and sociophoneticians, this tool automates phoneme-level annotation of Bangla speech data, streamlining acoustic analysis workflows. Currently, the toolkit runs on Linux commandline; Windows compatibiltiy and a GUI version is in development. Download Bangla-Align
Praat scripts
get_formants_for_NORM: This script extracts first and second formant measurements from audio data and exports them to TSV files formatted for compatibility with NORM, a vowel normalization toolkit.
extract_clips_by_labels_in_TG.praat: This script extracts audio segments and their corresponding TextGrid annotations based on specified labels within TextGrid files. The script processes all audio-TextGrid file pairs in a designated folder iteratively.
R scripts
measure_complexity_of_tongue_shape_contour: The script calculates quantitative complexity metrics for a given contour, implementing the approach from the Kabakoff et al. (2021)*. In particular, it computes two measures: 1) Modified Curvature Index (MCI): the total integrated absolute curvature along the contour (a measure of overall “wrinkliness”), and 2) Number of Inflections (NINFL): the count of curvature sign‐changes (after smoothing/thresholding), indicating how many times the contour bends in alternate directions. These metrics allow you to compare contour complexity across shapes in a principled, reproducible way.