Automated Creak Detection in Spanish Speakers with and without AdLD

Area of Science:

• Neurolinguistics
• Speech-language pathology
• Acoustic analysis

Background:

• Adductor laryngeal dystonia (AdLD) is a neurological voice disorder.
• Automated creak detection shows promise for AdLD in English speakers.
• Research on laryngeal dystonia (LD) tools in Spanish is limited.

Purpose of the Study:

• To determine if creak analysis can differentiate Spanish-speaking individuals with and without AdLD.
• To assess the diagnostic accuracy of creak detection in Spanish speakers with voice disorders.

Main Methods:

• Recorded 20 AdLD speakers, 20 controls, and 20 with glottic insufficiency.
• Participants read Spanish stimuli designed for LD screening.
• Used an open-source creak detector and calculated mean smoothed cepstral peak prominence (CPPS).

Main Results:

• Creak showed moderate correlation with CPPS (r = -0.56).
• Creak significantly differentiated between groups (F(3, 56) = 13.52).
• Creak accurately differentiated AdLD from controls (AUC = 0.88) and glottic insufficiency (AUC = 0.73).

Conclusions:

• Creak analysis moderately discriminates AdLD in Spanish speakers.
• Creak remains statistically significant between AdLD and controls/glottic insufficiency when controlling for CPPS.
• Further research is needed to establish creak's clinical utility for differential diagnosis in Spanish speakers.