The PSST Challenge is a shared task on computational analysis of disordered speech produced by aphasic individuals in a clinical context. Specifically, the challenge involves high-accuracy automatic phoneme recognition of naming tests, with particular attention paid to phonemic and semantic speech errors. Leveraging a widely-used library of aphasic speech, we will provide participants with a new set of high-quality phonemic transcripts and temporal annotations, making this an excellent opportunity for researchers from the wider machine learning community to begin engaging with an under-studied biomedical domain. We invite participation from novice and experienced ASR researchers alike, and will have funding for a limited number of travel assistance grants for students and postdocs.
For further details about the PSST Challenge, see further: description here including information on travel grants for students and postdocs.