app/metrics.py

# This module defines evaluation metrics

from yaml import warnings

warnings({"YAMLLoadWarning": False})

import panphon
import panphon.distance

ft = panphon.FeatureTable()
panphon_dist = panphon.distance.Distance()
inverse_double_weight_sum = 1 / (sum(ft.weights) * 2)


def per(prediction, ground_truth):
    """
    Phoneme Error Rate: the number of edits (substitutions, insertions, deletions)
    needed to transform the prediction into the ground truth divided by the length of the ground truth.
    """
    return panphon_dist.fast_levenshtein_distance(prediction, ground_truth) / len(
        ground_truth
    )


def fer(prediction, ground_truth):
    """
    Feature Error Rate: the edits weighted by their acoustic features summed up and divided by the length of the ground truth.
    """
    return (
        inverse_double_weight_sum
        * panphon_dist.weighted_feature_edit_distance(ground_truth, prediction)
        / len(ground_truth)
    )