Finding the sweet spot for phoneme in noise training

Reverse hierarchy theory proposes that rapid perception is based on high-level representations of the global, abstract categories of the perceived objects (Ahissar et al. 2009). Thus, in everyday communication and speech-in-noise understanding in particular, there is little time left for dwelling on fine spectro-temporal details of perceived speech. And yet, it is the fine details of speech sounds that are often ambiguous or unavailable to hearing impaired listeners and that would require repeated presentation. Phoneme training gives the listener the opportunity to concentrate on the detailed spectro-temporal representation, with repetition and feedback. However, the floor and ceiling effects of such training procedure diminish both the training benefit and the motivation of the listener. For each subject there is a narrow range of speech to noise ratios which will result in the largest training improvement. We shall call this the individual’s training sweet spot. In the study reported here, an automatic procedure for determining listener's training sweet spot was developed and tested. The procedure, training results and the subjective experiences of normal hearing and hearing impaired subjects will be presented.


Ahissar, Merav, et al. "Reverse hierarchies and sensory learning." Philosophical Transactions of the Royal Society of London B: Biological Sciences 364.1515 (2009): 285-299.