The rise of Alzheimer’s disease (AD) and behavioral variant frontotemporal dementia (bvFTD) calls for scalable approaches to identify and characterize patients. Automated speech analysis (ASA) is a low-cost innovation revealing early markers of dementia in many speech communities. Yet, no study has targeted Latin American Spanish speakers, exacerbating current gaps towards global approaches to dementia. This grant proposes a regional ASA framework to discover novel markers of AD and bvFTD in Spanish-speaking Latinos as a seed initiative to apply for and obtain larger funding. We will analyze pre-existing speech data from 30 AD patients, 30 bvFTD patients, and 30 healthy controls (HCs), combine ASA with machine learning to discriminate among groups via acoustic and/or linguistic metrics, and test whether these capture symptom severity in each disorder. Additional studies will be run in other groups, such as Parkinson’s disease (PD) patients.