Cortical dynamics of speech perception in adverse listening conditions

The perception of speech is usually an effortless and reliable process even in highly adverse listening conditions. In addition to external sound sources, the intelligibility of speech can be reduced by degradation of the structure of speech signal itself, for example by digital compression of sound. This kind of distortion may be even more detrimental to speech intelligibility than external distortion, given that the auditory system will not be able to utilize sound source-specific acoustic features, such as spatial location, to separate the distortion from the speech signal. The perceptual consequences of acoustic distortions on speech intelligibility have been extensively studied. However, the cortical mechanisms of speech perception in adverse listening conditions are not well known at present, particularly in situations where the speech signal itself is distorted. The aim of this thesis was to investigate the cortical mechanisms underlying speech perception in conditions where speech is less intelligible due to external distortion or as a result of digital compression. In the studies of this thesis, the intelligibility of speech was varied either by digital compression or addition of stochastic noise. Cortical activity related to the speech stimuli was measured using magnetoencephalography (MEG). The results indicated that degradation of speech sounds by digital compression enhanced the evoked responses originating from the auditory cortex, whereas addition of stochastic noise did not modulate the cortical responses. Furthermore, it was shown that if the distortion was presented continuously in the background, the transient activity of auditory cortex was delayed. On the perceptual level, digital compression reduced the comprehensibility of speech more than additive stochastic noise. In addition, it was also demonstrated that prior knowledge of speech content enhanced the intelligibility of distorted speech substantially, and this perceptual change was associated with an increase in cortical activity within several regions adjacent to auditory cortex. In conclusion, the results of this thesis show that the auditory cortex is very sensitive to the acoustic features of the distortion, while at later processing stages, several cortical areas reflect the intelligibility of speech. These findings suggest that the auditory system rapidly adapts to the variability of the auditory environment, and can efficiently utilize previous knowledge of speech content in deciphering acoustically degraded speech signals.