Human auditory cortex activity shows additive effects of spectral and spatial cues during speech segregation

• Published in: Cerebral cortex (New York, N.Y. 1991) N.Y. 1991), 2011-03, Vol.21 (3), p.698-707
• Main Authors: Du, Yi, He, Yu, Ross, Bernhard, Bardouille, Tim, Wu, Xihong, Li, Liang, Alain, Claude
• Format: Article
• Language: English
• Subjects: Adult; Auditory Cortex - physiology; Brain Mapping; Cues; Female; Humans; Magnetoencephalography; Male; Speech Perception - physiology; Young Adult
• ISSN: 1047-3211; 1460-2199
• DOI: 10.1093/cercor/bhq136

In noisy social gatherings, listeners perceptually integrate sounds originating from one person's voice (e.g., fundamental frequency (f(0)) and harmonics) at a particular location and segregate these from concurrent sounds of other talkers. Though increasing the spectral or the spatial distance between talkers promotes speech segregation, synergetic effects of spatial and spectral distances are less well understood. We studied how spectral and/or spatial distances between 2 simultaneously presented steady-state vowels contribute to perception and activation in auditory cortex using magnetoencephalography. Participants were more accurate in identifying both vowels when they differed in f(0) and location than when they differed in a single cue only or when they shared the same f(0) and location. The combined effect of f(0) and location differences closely matched the sum of single effects. The improvement in concurrent vowel identification coincided with an object-related negativity that peaked at about 140 ms after vowel onset. The combined effect of f(0) and location closely matched the sum of the single effects even though vowels with different f(0), location, or both generated different time courses of neuromagnetic activity. We propose that during auditory scene analysis, acoustic differences among the various sources are combined linearly to increase the perceptual distance between the co-occurring sound objects.