Multidimensional Acoustic-Prosodic Quantification Framework Using Unscripted Speech for Autism Spectrum Disorder Identification.

Researchers created a computer program that can help identify autism by analyzing how children speak during normal conversations. They studied 170 children aged 3-10 years and found that autistic children have different speech patterns - they speak less continuously, at different rates, and with different pitch and sound qualities. The program was 85% accurate at distinguishing autistic from non-autistic children. This could lead to easier autism screening without needing formal testing situations.

This study developed an automated framework to identify autism spectrum disorder (ASD) using unscripted speech analysis in children aged 3-10 years. Researchers analyzed spontaneous conversations from 88 children with ASD and 82 typically developing children during naturalistic interactions about daily topics. The framework extracted acoustic-prosodic features and achieved 85% accuracy in distinguishing ASD from typical development. Children with ASD showed distinct speech patterns including reduced speech continuity and rate, altered pitch characteristics, and different formant frequencies.

These patterns correlated with clinical language assessments and were more pronounced in open-ended dialogues compared to structured tasks, suggesting potential for large-scale ASD screening in non-clinical settings.

This technology could potentially enable earlier and more accessible ASD screening through analysis of everyday conversations. However, further validation in larger, diverse populations is needed before clinical implementation. The approach may be particularly valuable for identifying children who might not complete formal assessments.

Single study with relatively small sample size. No information provided about participant demographics, diagnostic methods, or comparison with existing screening tools. Unclear how the framework would perform across different languages, cultures, or clinical settings.

Although clinical observations have noted early speech abnormalities in children with autism spectrum disorder (ASD), automatic speech-based detection remains challenging. This is primarily due to the reliance on scripted tasks, which younger children often struggle to complete and which are not generalizable to large-scale, non-clinical screening. To address this, we developed an unscripted speech-based framework to quantify atypical acoustic-prosodic patterns for automatic ASD identification in naturalistic interactions. It processes free-flowing conversations, extracts multidimensional acoustic features from the time and frequency domains, and models ASD-related prosodic patterns for classification.

For evaluation, we collected spontaneous speech from 88 children with ASD (3-10 years) and 82 typically developing (TD) children (3-9 years) during naturalistic interactions on daily topics (e.g., toys, animated movies, storybook reading). Group comparisons revealed atypical prosodic patterns in ASD, including reduced speech continuity, speech rate, and Formant 3, alongside increased zero-crossing rate, pitch, pitch variability, and Formant 1 (all p < 0.01). Using these features, a linear discriminant analysis classifier achieved robust performance (accuracy = 0.85 ± 0.07, F1 = 0.86 ± 0.07). Further analyses indicated no significant gender interaction (p > 0.05), but a pronounced effect of speech context (p < 0.01), with atypical patterns being more evident in open-ended dialogues than in text-guided settings.

Moreover, these patterns correlated with clinical scores (p < 0.05), particularly language ability, demonstrating the framework's utility for assessing ASD severity. These findings underscore the importance of analyzing unscripted speech to capture atypical prosodic patterns and provide a basis for large-scale ASD screening outside clinical settings.