Alzheimer's disease is the most common neurodegenerative dementia in elderly people in the world and its diagnosis requires a wide medical evaluation, supported by cognitive tests, clinical and imaging exams. Identifying the disease through speech can reduce the cost and time of medical diagnosis. Emotional states are important performance indicators of cognitive processes. Intelligent and non-invasive computational techniques can become relevant support tools for an early medical diagnosis. Therefore, this article addresses the use of emotion recognition through voice as a biomarker to identify the presence of Alzheimer's disease. The proposed method is based on the extraction of emotional features from speech and pattern recognition using neural networks. The results of the experiments reached an accuracy of 72.61%, a precision of 72.90% and a recall of 72.50% through cross-validation of the data.

No abstract