Here we present speech recognition data in quiet and noise from thousands of patients with hearing loss. These data support using speech-in-noise testing as the default in audiologic assessment, reserving word-recognition in quiet for when performance is likely to be suoboptimal. We will also demonstrate between-session repeatability of speech tests as a function of hearing loss and baseline performance. Finally, we will show preliminary results from our AI-based tool for automatic speech recognition, which is capable of scoring audiologic tests in any language. Taken together, this research aims to modernize audiologic practice and better align testing with patient concerns.
Attendees will be able to:
- Describe the relationship between speech recognition in quiet and noise
- Articulate how repeatability of speech recognition tests change with increasing hearing loss
- Describe new developments in artificial intelligence for scoring tests of speech recognition in quiet and noise
