Date of Award
Doctor of Philosophy
Lisa Lucks Mendel
Eugene H Buder
Gavin M Bidelman
Although the AzBio test is well validated, has effective standardization data available, and is highly recommended for Cochlear Implant (CI) evaluation, no attempt has been made to derive a Frequency Importance Function (FIF) for its stimuli. In the first phase of this dissertation, we derived FIFs for the AzBio sentence lists using listeners with normal hearing. Traditional procedures described in studies by Studebaker and Sherbecoe (1991) were applied for this purpose. Fifteen participants with normal hearing listened to a large number of AzBio sentences that were high- and low-pass filtered under speech-spectrum shaped noise at various signal-to-noise ratios. Frequency weights for the AzBio sentences were greatest in the 1.5 to 2 kHz frequency regions as is the case with other speech materials. A cross-procedure comparison was conducted between the traditional procedure (Studebaker and Sherbecoe, 1991) and the nonlinear optimization procedure (Kates, 2013). Consecutive data analyses provided speech recognition scores for the AzBio sentences in relation to the Speech Intelligibility Index (SII). Our findings provided empirically derived FIFs for the AzBio test that can be used for future studies. It is anticipated that the accuracy of predicting SIIs for CI patients will be improved when using these derived FIFs for the AzBio test. In the second study, the SIIfor CIrecipients was calculated to investigate whether the SII is an effective tool for predicting speech perception performance in a CI population. A total of fifteen CI adults participated. The FIFs obtained from the first study were used to compute the SII in these CI listeners. The obtained SIIs were compared with predicted SIIs using a transfer function curve derived from the first study. Due to the considerably poor hearing and large individual variability in performance in the CI population, the SII failed to predict speech perception performance for this CI group. Other predictive factors that have been associated with speech perception performance were also examined using a multiple regression analysis. Gap detection thresholds and duration of deafness were found to be significant predictive factors. These predictor factors and SIIs are discussed in relation to speech perception performance in CI users.
dissertation or thesis originally submitted to the local University of Memphis Electronic Theses & dissertation (ETD) Repository.
Lee, Sung Min, "Predicting Speech Recognition using the Speech Intelligibility Index (SII) for Cochlear Implant Users and Listeners with Normal Hearing" (2017). Electronic Theses and Dissertations. 1745.