Office of Technology Transfer – University of Michigan

Smartphone app for aphasia therapy

Technology #6294

After a stroke, the road to recovery is anything but simple. In nearly 40 percent of cases – as well as in other instances of brain trauma – patients will develop a speech-language disorder called aphasia. This condition currently hinders the ability of over 1 million Americans to understand and be understood in both written and verbal communication. Although intensive therapy has been proven to substantially alleviate symptoms, treatment is not feasible investment of time or money for many. Building off recent studies which have demonstrated the utility of computer software for aphasia therapy, this technology describes a powerful smartphone app capable of assessing patient speech patterns without a prior sample and in noisy environments – a must for any rehabilitation software intended for use on a mobile device. Further, through a trimodal alignment method, the algorithm is capable of assessing the patient’s speech on several levels and providing the appropriate feedback through video, diagrams etc. to enhance patient understanding and compliance.

Smart rehabilitation on a mobile device

Unconstrained speech recognition is generally challenging. Thus, this technology relies on context-based keyword spotting to improve accuracy based on the prompt supplied to the patient. Upon acquisition, the aphasic speech sample is compared to health templates across three acoustic levels (word, syllable, and/or phone). By matching across these three dimensions, a more accurate assessment of speech quality can be made and – more importantly – issues in need of correction can be more readily identified. Since feedback in the form of mouth movement or tongue placement diagrams is generally better received by aphasic patients, such measures are accordingly incorporated into the technology.


  • Supplementary practice for individual speech therapy
  • In-home, low-cost, alternative to speech therapy sessions


  • Reliable speech quality assessment
  • Concrete, actionable phonological feedback
  • Personalization motivates users and increases practice efficacy