Adductor laryngeal dystonia (AdLD) is a rare neurological disorder characterized by spasms of the laryngeal muscles, causing sporadic irregularities in voicing. These acoustic behaviors are classified as: creak, or sudden changes in vocal timbre; and phonatory breaks, or inappropriate breaks in voicing. Presently, assessment of AdLD is subjective and misdiagnosis is common. As a foundation for objective assessment methods, we characterized the laryngeal and acoustic behaviors of speakers with AdLD (n=16, 63.8±8.9 years) paired with age-and-sex-matched speakers with typical voices (n=16; 64.8±9.4 years). High-speed videoendoscopy (HSV) was used to visualize the laryngeal behaviors empirically observed in AdLD: laryngeal compression, where the vocal folds are compressed by surrounding structures; and laryngeal obstruction, where the vocal folds are obstructed by the epiglottis or arytenoids. Speech was recorded using a headset microphone simultaneously with HSV. 

We found acoustic discontinuities to be more prevalent in speech samples from AdLD speakers (p=0.04, d=0.88). Specifically, speech samples from AdLD speakers contained 13.6% creak, compared to 7.6% in control speaker samples. Further, we found laryngeal obstructions to be significantly more prevalent in speakers with AdLD(p=.01, d=1.02). The percentage of laryngeal compression in both typical and AdLD speakers was highly varied; there was no significant difference between group means. 

Characterizing laryngeal and acoustic behaviors will inform and expedite AdLD assessment. Paired laryngeal and acoustic data collected in this study may serve to support the development and validation of non-invasive, automated AdLD detection.