This research focused on addressing the common occurrence of depression in individuals with Parkinson's Disease (PD), a neurodegenerative disorder. Depression can significantly affect a person's functioning, making early detection crucial for effective treatment. The analysis explored the use of voice recordings from PD patients to extract paralinguistic features, which are non-verbal elements of speech such as tone, pitch, and rhythm. These features were then utilized to train Machine Learning and Deep Learning models with the objective of predicting depression. The results of the research revealed promising outcomes, with the models achieving accuracies as high as 0.77 in accurately classifying subjects as depressed or non-depressed. These findings suggest that voice recordings can serve as digital biomarkers to screen for depression among PD patients. By leveraging these paralinguistic features, healthcare professionals could potentially identify depression in PD patients at an earlier stage, facilitating prompt intervention and enhancing treatment outcomes. The implications of this research are as follows. Implementing voice-based screening tools could offer a non-invasive and easily accessible method to assess the mental well-being of PD patients. Such early detection could help clinicians to tailor treatment plans accordingly, ensuring that patients receive appropriate care for both PD and comorbid depression. Ultimately, the integration of voice-based screening into routine clinical practice has the potential to improve the overall quality of patients with PD, leading to better mental health outcomes.