Cortical Representations of Speech in a Multitalker Auditory Scene

Area of Science:

• Neuroscience
• Auditory Perception
• Human Auditory Cortex

Background:

• Auditory scene analysis relies on a hierarchical auditory system.
• Peripheral auditory nerve representations evolve into distinct auditory-object-based representations in the auditory cortex.
• Understanding how complex acoustic scenes are processed through auditory cortical hierarchy is crucial.

Purpose of the Study:

• Investigate the hierarchical representation of complex acoustic scenes with multiple speech sources in the human auditory cortex.
• Determine how attended and unattended speech streams are processed at different hierarchical stages.
• Characterize the neural representations of auditory objects across the auditory hierarchy.

Main Methods:

• Magnetoencephalography (MEG) recordings from human listeners.
• Simulated cocktail party environment with multiple speech sources.
• Systems-theoretic methods of stimulus reconstruction.

Main Results:

• Primary-like auditory cortical areas show spectrotemporal-based representations of the entire auditory scene, with attended and ignored speech equally represented.
• Higher-order auditory cortical areas represent attended speech with higher fidelity than unattended speech.
• Unattended background streams are represented as a single object, not segregated objects.

Conclusions:

• Demonstrates a progression of auditory scene representation through the human auditory cortex hierarchy.
• Early auditory areas provide a global spectrotemporal overview, while later areas enable selective attention.
• The auditory system effectively segregates attended speech from background noise at higher processing levels.