August 2023. Presented paper on a phonetic methods survey at the International Congress of Phonetic Sciences.
Vowel Clustering Methods GitHub
This project systematically compares two clustering algorithms: K-means and OPTICS (similar to DBSCAN) on both theoretical and empirical levels. The project finds that K-means entails assumptions that both theoretically and empirically pose issues for meaningful interpretations of the results. OPTICS is proposed as an alternative that has potential for learning meaningful patterns in exploratory vowel space analysis.
The slides and code are available at this GitHub repository.