Parkinson’s disease (PD) is a neurodegenerative disorder that affects a significant number of people worldwide. Despite advances in understanding this disease, its diagnosis remains a significant challenge. This is mainly because physicians base their assessment on clinical history and a review of manifested signs and symptoms. However, this assessment can be complicated due to the heterogeneity and complexity of PD characteristics. Given these complexities, much of the research has focused on discovering new early biomarkers of this disease; one of them is voice, whose alterations often appear in the early stages of the disease and even in prodromal phases in around 90% of patients diagnosed with the disease. This paper highlights the use of machine learning techniques that, through acoustic features of voice, can detect the presence or absence of this pathology in its early stages. The employed models include Boosting, Bagging, and Random Forest. The dataset used for the experiments was the Parkinson Speech Dataset with Multiple Types of Sound Recordings. The comparison of classification results by the models shows that the Boosting classifier is a robust machine-learning technique for detecting PD. This classifier model has a detection accuracy of 0.7340, a Precision of 0.7574, a sensitivity of 0.7529, a specificity of 0.7113, an F1 Score of 0.7552, and an AUC-ROC of 0.7758. Finally, it is concluded that machine learning approaches have the potential to provide physicians with additional tools to detect or diagnose PD accurately.