Automated Creak Identifies Laryngeal Dystonia During Conversational Speech

Area of Science:

• Laryngology
• Acoustic analysis
• Speech science

Background:

• Creak, an acoustic feature of voice, reflects fundamental frequency irregularities.
• Differentiating voice disorders like adductor laryngeal dystonia (AdLD) and muscle tension dysphonia (pMTD) is crucial for diagnosis.

Purpose of the Study:

• To evaluate if automated creak detection can differentiate speakers with AdLD, pMTD, and controls.
• To assess creak's discriminative ability during conversational speech.

Main Methods:

• 50 speakers each with AdLD, pMTD, and 50 controls produced reading and conversational speech.
• An automated creak detector quantified the percentage of creak.
• Receiver operating characteristic curve analyses determined discriminative ability.

Main Results:

• A significant effect of speaker group on creak percentage was found.
• Automated creak distinguished AdLD, pMTD, and control speakers in both speech tasks.
• Area under the curve values showed acceptable ability to differentiate AdLD speakers.

Conclusions:

• Automated creak analysis is a promising tool for distinguishing between AdLD, pMTD, and healthy speakers.
• This acoustic feature holds potential for clinical application in voice disorder identification.