The presence of prosodic anomalies in autistic children is recognized by experienced clinicians but their quantitative analysis is a cumbersome task beyond the scope of typical pen and pencil assessment. Quantifiable speech measurements, i.e., loudness and pitch, are correlated to prosody making them potential references for assisting the analysis. However, their importance is underestimated in the application of autism diagnosis. These measurements are largely ignored in mainstream autism diagnosis batteries and commonly replaced with few vague assessments. This research work emphasizes the potential of speech measurements on autism diagnosis through creating and evaluating a fully automatic speech-oriented autism detecting system. We propose an automatic approach allowing to tease apart various aspects of prosodic abnormalities and to translate them into fine-grained, automated, and quantifiable measurements.