Vocal and Facial Behavior During Affect Production in Autism Spectrum Disorder

Area of Science:

• Speech-language pathology
• Developmental psychology
• Biomedical engineering

Background:

• Autism spectrum disorder (ASD) presents challenges in social communication and emotional expression.
• Objective measurement of affective behaviors in children with ASD is crucial for diagnosis and intervention.
• Automated analysis of audiovisual data offers a novel approach to quantify subtle differences in emotional communication.

Purpose of the Study:

• To determine if automated audiovisual metrics from an affect production task can distinguish children with autism spectrum disorder (ASD) from typically developing controls.
• To investigate differences in vocal and facial emotional expression between children with ASD and neurotypical controls.

Main Methods:

• Forty children with ASD and 21 neurotypical controls interacted with a virtual agent, Tina.
• Participants communicated four emotions (happy, angry, sad, afraid) under varying task demands.
• Automated analysis of vocal (fundamental frequency, intensity, voice quality) and facial (lip, eye, eyebrow movements) metrics was performed.

Main Results:

• Children with ASD showed greater voice pitch variability and earlier pitch contour changes compared to controls.
• Significant differences in speech intensity and voice quality were observed under specific conditions.
• Facial metrics, including lower lip acceleration, lip width, eye opening, and eyebrow displacement, were key discriminators.
• Both speech and facial metrics achieved high classification accuracy above chance.

Conclusions:

• Automated speech and facial metrics derived from affect production tasks are effective in distinguishing children with ASD from neurotypical controls.
• These objective measures provide valuable insights into the affective communication differences in ASD.
• This approach holds potential for enhancing diagnostic tools and understanding emotional expression in ASD.