Children with Cleft Lip and Palate (CLP) may experience difficulties in oral communication, leading to other developmental problems such as delayed language acquisition and poor social skills; thus, early treatment is essential for successful speech rehabilitation. In this paper, we propose a methodology for automatically assessing the phonological precision of children with CLP. We propose to use the probabilities obtained from a phonological class recognizer to measure phonological precision during connected speech. Furthermore, we compute the nasal-to-sound ratio to improve the automatic detection of the nasality level. For this, we considered speech recordings of 88 children with CLP, assessed by a clinician according to four nasality levels: normal, mild, moderate, and severe. We obtained an F1-score of up to 0.54 for detecting the nasality level automatically. The results suggest that phonological analysis can be used for individualized speech rehabilitation.
Cite as: Arias-Vergara, T., Londoño-Mora, E., Pérez-Toro, P.A., Schuster, M., Nöth, E., Orozco-Arroyave, J.R., Maier, A. (2023) Measuring Phonological Precision in Children with Cleft Lip and Palate. Proc. INTERSPEECH 2023, 4638-4642, doi: 10.21437/Interspeech.2023-2099
@inproceedings{ariasvergara23_interspeech, author={Tomás Arias-Vergara and Elizabeth Londoño-Mora and Paula A. Pérez-Toro and Maria Schuster and Elmar Nöth and Juan Rafael Orozco-Arroyave and Andreas Maier}, title={{Measuring Phonological Precision in Children with Cleft Lip and Palate}}, year=2023, booktitle={Proc. INTERSPEECH 2023}, pages={4638--4642}, doi={10.21437/Interspeech.2023-2099} }