Page 31 - 2017Spring
P. 31

received speech envelope, degrading the speech modulation transfer. Similarly, if temporal dips in the speech envelope fall below the threshold of audibility in a frequency band, then intelligibility will be reduced.
A drawback of the AI, SII, and STI approaches is that any nonlinear aspects of processing by the human auditory sys- tem and degradations due to hearing loss can only be imple- mented in an ad hoc manner. Furthermore, it is now known that hearing impairment can occur without being reflected in the audiogram (i.e., the clinical measure of behavioral thresholds for hearing). Animal studies suggest that the loss of approximately 50-90% of auditory nerve (AN) fibers can occur without affecting the audiogram, and moderate sound exposures have the potential to cause such neural degenera- tion without producing any damage to the hair cells of the cochlea that determine behavioral thresholds (Liberman, 2016). However, because of the highly stochastic nature of AN responses at the single-fiber level, the loss of AN fibers is expected to degrade the representation of suprathreshold sounds such as speech (Lopez-Poveda and Barrios, 2013). These limitations in the acoustic-based metrics restrict their usefulness in understanding and quantifying how people perceive speech and, consequently, their utility for evaluat- ing and improving devices such as hearing aids and cochlear implants. This has motivated the development of physiologi- cally based intelligibility predictors by a number of different research groups.
Anatomy and Physiology of the Ear
Before jumping into the details of the physiological predic- tors, it is helpful to review the underlying auditory anatomy and physiology. Figure 1 shows the anatomy of the human ear. The pinna and ear canal of the outer ear funnel sounds to the middle ear, which consists of the tympanic membrane (eardrum) and three tiny bones: the malleus, incus, and sta- pes. The footplate of the stapes lies atop the oval window, a membranous opening to the fluid-filled bony cochlea of the inner ear. The difference in size between the tympanic mem- brane and the oval window as well as the lever action of the middle ear bones produces a fairly good acoustic impedance match between the air in the ear canal and the fluids within the cochlea.
Figure 2 shows a cross section of the sensory structure in the cochlea, the organ of Corti, and Figure 3 depicts how the acoustic vibrations conveyed to the oval window by the middle ear are transduced into neural “spikes” in the AN
Figure 1. Anatomy of the human ear. The pinna and external audi- tory canal form the outer ear. The middle ear consists of the tympanic membrane and the three tiny bones: the malleus, incus, and stapes. The cochlea is the sensory receptor organ of the inner ear, and the au- ditory nerve conveys the transduced acoustic information to the audi- tory pathways of the brain. Adapted from Chittka and Brockmann (2005) under the terms of the Creative Commons Attribution License © 2005 Chittka and Brockmann.
Figure 2. Cross section of the organ of Corti in the cochlea. The basi- lar membrane (BM) spans the length of the fluid-filled cochlea, and its mechanical properties cause it to be tuned to different acoustic fre- quencies (see Figure 3, bottom). The organ of Corti, containing the sensory receptor cells known as inner hair cells (IHCs) and outer hair cells (OHCs), sits on top of the basilar membrane. Vertical vibration of the BM causes a shearing force to be applied to the hairlike cilia of the IHCs and OHCs, which, in turn, generates a transduction current and a subsequent change in the electrical potentials within the hair cells. The OHCs have an electromotile response (as described in the article by Brownell in this issue of Acoustics Today) that leads to a time-varying, nonlinear BM vibration pattern. The IHCs release a neurotransmitter that generates information-carrying “spikes” in the electrical potential of auditory nerve fibers. Reprinted with permis- sion from Brownell (1997).
   Spring 2017 | Acoustics Today | 29

   29   30   31   32   33