Objectives: To conduct preliminary tests of whether early infant cry acoustics differentiate infants at high and low risk for autism.

Methods: Vocal samples were recorded at home in 6-month-old infants at risk for autism (younger siblings of children with ASD; n= 21) and low risk infants (n= 18). Standard definitions were used to identify cries and fusses. Samples with adequate recording quality were selected for analysis. Samples were excerpted to audio tape and processed by cry analysis software. Samples were filtered (5 kHz), digitized (10 kHz), and separated by utterance. The log magnitude spectrum for each 25-ms utterance block was computed (using a Fast Fourier Transform).

Results: Vocalizations of high risk infants showed less variability in energy and greater variability in fundamental frequency (pitch). Two children had confirmed autism diagnoses at age 3; one had vocalizations that could be analyzed. Z-scores were calculated to compare this child's scores to the control group. This child's cry had low energy (Z= -3.0), high fundamental frequency/pitch (Z= +1.4), and high dysphonation (Z= +2.4). This pattern was not fully observed in other infants. One low risk infant had low energy and high dysphonation, but differences were smaller and there were no differences in pitch. One high risk infant had a similar pattern of low energy and high dysphonation, but did not differ in pitch.

Conclusions: Infant cry may be one early indicator of disrupted vocal development in autism. These findings raise the hypothesis that autism may be associated with disruptions in the coordination of vocal production during infancy.