Parkinson’s disease (PD) is a neurodegenerative disorder characterized by progressive deterioration of motor functions and alteration of voice characteristics; thus, the earlier this disorder is identified, the better the life quality for the affected individual. Recent developments in the field of artificial intelligence have pointed out speech as a non-invasive, low-cost biomarker for Parkinson’s detection. In the present study, we investigate the performance of two different models, MLP-Mixer and Support Vector Machines (SVM), on an automated PD detection task using voice data. Leveraging a UCI dataset comprising voice samples from PD patients and healthy individuals, we extract Mel-spectrogram features that capture time-frequency patterns indicative of Parkinson’s disease . Each model is separately trained and evaluated based on these features to classify voice samples as either PD-positive or healthy. The obtained accuracy results of our research is 92%, highlighting the ability of MLP-Mixer model to serve as a low-cost, easily accessible assistive model toward the diagnosis of PD.