New methods for identifying concussions are being researched in an attempt to find inexpensive, more reliable protocols. Importantly, there is a general lack of understanding regarding the effect of concussion on speech production and the voice of brain injured individuals especially those who have sustained a concussion [1]. Additionally, there is even less research available comparing the speech and voice characteristics of concussed athletes to that of normal healthy speakers.

Few researchers have attempted to analyze the voices of concussed athletes [2]. Those that have to date examined the voices of these athletes suggest that changes in the voice following head injury may be used to identify individuals with concussion. 
More robust voice analysis, less affected by recording noise and speech variability is needed. The current research proposes and outlines the pilot research involving formant analysis of athletes following concussion. Are there measurable differences in the  formant frequencies of normally healthy athletes versus athletes who have a concussion. Participants consisted of 92 athletes from throughout the Midwest. Several formant analyses were performed on vowels automatically extracted from connected speech of concussed athletes before and after sustaining a concussion. Formant frequencies were then extracted from these vowels and vowel space area calculations were performed [3]. F1-F2 trajectories can potentially predict articulatory dynamic specific to pre- and post-concussion. F2 predicted concussion status: paired t-test (t(83) = -2.0580, p = 0.04 (<0.05)). F1-F2 trajectories suggest lack of control over tongue movements during vowel utterance after concussion.  Types of discriminative features vary among individuals. Separation of concussed voice from non-concussed one requires investigating a multidimensional acoustic space

Application of Automated Vowel Space Area Analysis to Predict Concussion Status