Speech, as one of the most complex human functions, involves over a hundred muscles and multiple brain areas, making it a valuable marker sensitive to neurodegenerative changes. With the advent of modern technologies, the link between speech disruption and neurodegeneration is gaining increasing research attention. The inexpensive and noninvasive nature of speech assessment, coupled with the broad availability of devices like smartphones, enables detailed speech monitoring during routine examinations or even in patients’ daily lives. However, current knowledge of speech impairment largely relies on cross-sectional studies, lacking insights into individual speech changes associated with disease progression. Most studies have been conducted on clinically diagnosed PD cohorts, and the disease’s heterogeneity, along with potentially non-linear speech deterioration, may limit the findings’ applicability to the prodromal stage. To effectively study PD in its prodromal phase, large-scale, population-based prospective studies are necessary. However, with PD’s 2% lifetime prevalence, designing and conducting such studies is challenging. To limit the study’s scope, one approach is to preselect participants at higher risk for developing the disease, thereby reducing the study to a smaller, high-risk cohort of participants with known disease markers. Over the past decade, studies on automatic speech analysis in RBD patients enabled the direct characterization of speech patterns indicative of prodromal PD. This review synthesizes current research on automated speech analysis in individuals with RBD, exploring its potential as an early biomarker for neurodegenerative disease progression. We systematically reviewed studies published between 2014 and 2024, focusing on methodologies employing computational acoustic analysis.