Cortical tracking of speakers' spectral changes predicts selective listening

Area of Science:

• Neuroscience
• Auditory Perception
• Speech Processing

Background:

• Distinguishing speakers in social settings is crucial for communication.
• The neural mechanisms for processing vocal characteristics, especially in complex auditory scenes, remain unclear.
• Understanding how the brain decodes vocal information is key to auditory scene analysis.

Purpose of the Study:

• To investigate the neural basis of tracking slow frequency modulations in speech.
• To determine if the brain's tracking of vocal spectral information is linked to selective listening abilities.
• To explore how cortical activity relates to identifying individual speakers in complex auditory environments.

Main Methods:

• Utilized electroencephalography (EEG) to measure neural responses to speech.
• Employed single-trial neural tracking and stimulus reconstruction methods.
• Assessed participants' performance on a cocktail party comprehension task with single and mixed speakers.

Main Results:

• Demonstrated robust neural tracking of slow (delta-theta) frequency modulations in the 3.5-5 KHz range.
• Showed that spectral spacing information (ΔF) from vocal tract resonances is decodable from neural signals.
• Found that EEG-based spectral tracking ability predicted success in selective listening tasks with competing speech.

Conclusions:

• The communicating brain locks cortical rhythms to upper vocal tract resonances for real-time speaker identification.
• Neural tracking of vocal spectral contours is a fundamental mechanism for auditory scene analysis.
• This tracking ability is essential for listeners to selectively attend to and comprehend speech in noisy conditions.