A wide variety of disorders known as autism, or autism spectrum disorder (ASD), are characterized by difficulties with speech, nonverbal communication, social skills, and repetitive activities. This developmental disability is brought on through variations in the brain. Instead of relying on a child’s intellectual and behavior capacity, which takes time, Magnetic Resonance Imaging (MRI) regarding the brain is used for autism diagnosis. The successful implementation of early intervention methods depends on the early detection of ASD. To this end, we have developed a ML model which uses the sMRI characteristics and have improved the accuracy related to the results with the use of such proposed method, which is an ensemble soft-voting classifier combining 5 algorithms of ML. In a case when tested on ABIDE datasets, our system demonstrated excellent accuracy (96.21%), precision (99.1%), recall (96.4%), and F1-score (93.9%).