Seminario INCyL – Robert Lutfi

Difficulty recognizing speech in noise is the number one problem for those with hearing loss. Yet, the degree
of difficulty among individuals differs widely unrelated to the single measure that is the clinical gold standard for
evaluating hearing, the audiogram. Research has isolated a host of factors contributing to these differences,
both auditory and nonauditory, but these factors do not operate in isolation. In this talk I will describe work in
our lab using integrative modeling to quantify the relative impact and interaction of factors affecting speech
recognition in noise for individual hearing-impaired adults. We identify three main sources of variance in the
individual listener data with three generic stages of auditory processing: transduction, selective attention, and
decision. The partitioning of variance is done in a way that the relative impact at each stage can be quantified
as a measure of processing efficiency h universally comparable across different listening scenarios (Green and
Swets, 1966). New results using this approach suggest: (1) The overwhelming source of individual differences
in speech-in-noise recognition performance for both normal-hearing (NH) and hearing-impaired (HI) adults are
differences in decision noise associated with speech separation. Group differences between NH and HI
listeners in speech recognition in noise are due to differences in cue sensitivity. (2) For speech separation, HI
listeners place far greater reliance than NH listeners on the one cue for which they are most sensitive. HI
listeners, but not NH listeners, benefit by shifting diagnostic information to this most relied on cue. (3) A simple
to administer, 12-minute, speech-in-noise test can be used to estimate the listener-specific collective effect of
decision noise and cue reliance on the processing of linguistic, acoustic, and statistical properties of speech
[Work supported by NIH R01 DC001262-32].