Annually, 15 million people worldwide suffer from stroke leaving about one third permanently disabled. Speech impairment is a frequent symptom related to stroke, but distinguishing patients with subjective speech impairments (slurred speech, dysarthria, etc.) from those without underlying acute ischemic brain damage on clinical grounds represents a challenging problem in the emergency setting. In this project we thus aim to take advantage of the power of computational voice analysis and state-of-the-art pattern recognition methodology to contribute to a reliable diagnosis of speech impairments related to acute cerebral ischemia. We shall develop a tool to objectively analyse a patient’s voice characteristics for a time/cost-saving, easily applicable assessment in an everyday clinical setting. The underlying model will be implemented on the basis of voice recordings of patients with subjective speech impairments suspected to have an ischemic brainstem stroke and a matched control group. Standard neurological assessments and brain MRI will be applied to verify the “ground truth”.