2020 INTERSPEECH INTERSPEECH 2020

Whistled Vowel Identification by French Listeners

Abstract

In this paper, we analyzed whistled vowel categorization by native French listeners. Whistled speech, a natural, yet modified register of speech, is used here as a tool to investigate perceptual processes in languages. We focused on four whistled vowels: /i, e, a, o/. After a detailed description of the vowels, we built and ran a behavioral experiment in which we asked native French speakers to categorize whistled vowel stimuli in which we introduced intra- and inter- production variations. In addition, half of the participants performed the experiment in person (at the laboratory) while the other half participated online, allowing us to evaluate the impact of the testing set up. Our results confirm that the categorization rate of whistled vowels is above chance. They reveal significant differences in performance for different vowels and suggest an influence of certain acoustic parameters from the whistlers’ vowel range on categorization. Moreover, no effect or interaction was found for testing location and circumstances in our data set. This study confirms that whistled stimuli are a useful tool for studying how listeners process modified speech and which parameters impact sound categorization.

🧭 Keyword Pioneer — whistled vowel
🐝 Cross-Pollinator — Artificial Intelligence, Computer Science, Computer Vision, Data Science & Analytics, Deep Learning, Healthcare & Medicine, Interdisciplinary, Machine Learning, Mathematics & Optimization, Natural Language Processing, Speech & Audio