Neural correlates of speech quality dimensions analyzed using electroencephalography (EEG)

Area of Science:

• Neuroscience
• Auditory Perception
• Speech Processing

Background:

• Speech quality is perceptually decomposed into discontinuity, noisiness, and coloration.
• Previous electroencephalography (EEG) studies linked perceived intensity of speech quality dimensions to brain activity.
• Neurophysiological dissociability of these speech quality dimensions remained uninvestigated.

Purpose of the Study:

• To investigate if speech quality dimensions (discontinuity, noisiness, coloration) are dissociable on a neurophysiological level.
• To examine the electro-electrical correlates of speech quality perception using EEG.
• To determine if EEG can differentiate speech degradations based on specific quality dimensions.

Main Methods:

• High-quality speech stimuli were degraded along single dimensions (discontinuity, noisiness, coloration) to equal perceived degradation intensity.
• Participants performed a three-stimulus oddball task with standard and infrequent oddball stimuli.
• EEG recorded brain activity, focusing on the P300 event-related potential (ERP) component and its subcomponents (P3a, P3b).

Main Results:

• Significant modulations in P300 amplitude and latency were observed based on the perceptual dimensions of speech quality.
• Both standard and oddball stimuli demonstrated these modulations, indicating sensitivity to quality dimensions.
• The findings suggest distinct neurophysiological responses to different speech quality attributes.

Conclusions:

• EEG is a valuable tool for dissociating speech degradations based on specific perceptual dimensions, not just overall intensity.
• Neurophysiological methods can differentiate the distinct quality dimensions of speech.
• This research provides a foundation for understanding the neural basis of speech quality perception.