Automatic Acoustic Workflow for Speech: Testing Usability in a Research Lab

Abstract

Record ID: 183

Type: Poster Presentation (in-person)

Advisor: Carrie Rountrey

Abstract: Speech-Language Hearing (SLH) scientists and clinicians are interested in acoustic analysis to reflect physiological states in persons with and without speech disorders. Some acoustic measures correspond with how well people are understood by their conversational partners; this is referred to as intelligibility. Analyzing large sound files representative of intelligibility in connected speech is time-consuming and impractical for clinicians. Opportunities exist to leverage technology to automatically analyze these sound files. The CINCI-lab designed an automatic acoustic workflow (AAW) in collaboration with computer science. This new AAW needs a usability assessment. AAW is a computerized tool that analyzes speech sound files for acoustic measures associated with intelligibility. Usability refers to the ease of use and effectiveness of a specific tool; for this project, AAW. We aim to investigate the usability of AAW, so SLH students can reliably use this tool for sound file analysis. Our framework for establishing usability is as follows: Learnability: The level of difficulty to complete the task at hand using the workflow. Satisfaction: Measuring the level of future engagement of AAW through self-evaluation. Error tolerance: How efficient AAW is in notifying errors and the user's ability to recover the error. Teachability/Implementation: Effectiveness with which we can teach AAW to another lab member and reasonably expect implementation in our own lab. We will present our findings on AAW usability in terms of learnability, satisfaction, error tolerance, and teachability/implementation. We will also deliver recommendations to make AAW more user-friendly as needed for SLH scientists.