Neurodegenerative diseases (NDD) are a vast cluster of neurological disorders that debilitate patients’ cognitive health and physical abilities, leading to a decline in autonomy and quality of life. They affect 15% of the global population and typically have an unexplained onset and insidious course. The most prevalent neurodegenerative diseases are Parkinson’s disease (PD), amyotrophic lateral sclerosis (ALS), and Alzheimer’s disease (AD). Despite their prevalence, there are few disease-modifying therapies available to prevent or treat them. Therefore, identifying possible voice-related biomarkers to diagnose and monitor these pathologies would be highly important and useful. There are several studies that show promising results regarding possible voice-related digital biomarkers in NDD: traditional (perturbation) and cepstral/spectral acoustic measures, but also new technologies, such as artificial intelligence, machine learning, and deep learning (automatic voice analyzer and speech recognition systems). All these measures may be useful to perform voice analysis and identify subtle preclinical voice impairments in neurodegenerative diseases, but we need large clinical trials to ensure their accuracy and reliability.