Implementation and automation of natural speech in aphasia

Approximately one third of stroke survivors is affected by a language disorder (aphasia). Aphasia can be expressed through difficulties in both production and comprehension of spoken and/or written language and thus influence everyday life and communication. Natural speech analysis in aphasia is gaining popularity since classical aphasia assessment includes tasks that are not representative for daily communication, resulting in progress on this assessment not being perceived as progress in daily life by the person with aphasia and the environment. Furthermore, natural speech analysis can provide essential information for both direct therapy with the person with aphasia as well as for indirect therapy involving the communication partner. There are currently few guidelines for clinicians and researchers regarding transcribing and analyzing natural speech. Moreover, this is time consuming and natural speech analysis is therefore only limited applied in clinical practice, despite previous research emphasizing the importance of natural speech analysis as part of aphasia rehabilitation. This study aims to lower the workload of natural speech analysis in persons with aphasia and their communication partner by validating automatic speech recognition and assessment and implementing these techniques in clinical practice.