Age Classification: Comparison of Human vs Machine in Prompted and Spontaneous Speech

Authors: Felix Burkhardt, Markus Brückl, Björn W. Schuller

Abstract:

We report on the results of an investigation to classify speaker age in vocal utterances with state-of-the-art machine learning algorithms on a small data set. We compare results of manual measurement, i. e., supervised automated extraction of phonetically interpretable measures and observation (by hypothesis tests via procedures like regression analysis) with the outcomes of experiments based on recent machine learning. On isolated vowels the machine outperformed the human estimates.