Determine enrollment scaling across dysarthria severities and aetiologies

Characterize how the relationship between patient enrollment-audio duration and dysarthric automatic speech recognition performance varies across dysarthria severities and aetiologies beyond the single severe post-stroke Hungarian speaker evaluated in the study.

Background

The enrollment-time experiment measures CER at 1, 3, 5, 10, 15, and 30 minutes using one severe post-stroke speaker speaking Hungarian. It finds continued gains with additional enrollment audio but does not establish whether the observed curve applies to patients with other severities or causes of dysarthria.

The authors explicitly identify the transfer of this enrollment curve to other clinical populations as unresolved and defer it to follow-up work.

References

The shape of the curve at other severities and aetiologies is a question we defer to follow-up work (Section~\ref{sec:conclusion}).

— Choosing a PEFT Variant for Per-Patient Dysarthric ASR: A Single-Speaker Case Study on Two ASR Bases  (2609.02735 - Muller et al., 2 Sep 2026) in Discussion, subsection “Clinical deployment: enrollment audio”