Musical instrument quality is typically tested using double-blind listening tests to avoid bias associated with superficial characteristics such as age and make. In this study, we propose a new method that uses electroencephalography (EEG) and machine learning to examine the possible separation of acoustic and electric violins from brain data. The experiment involved participants listen to melodies expressing different emotions and rate anonymous violin preferences while EEG was recorded. The analysis algorithm comprised a Riemannian-based common spatial pattern (CSP) spatial filtering method and a logistic regression (LR) classifier. Three experiments were then completed. First, a survey to confirm the expressed emotion of each audio clip was conducted. Second, a replication of previous EEG tests was done to classify valence and arousal. The results showed classification accuracies of 70.71% and 69.44% for valence and arousal, respectively. Third, a violin type classification with an accuracy of 61.56% was accomplished by pairing EEG data of the same melody with both violins into the same feature vector. This paper demonstrates a proof of concept for violin type classification using machine learning algorithms.