Regression between EEG and Speech Signals for Spoken Vowels

Area of Science:

• Neuroscience
• Speech Science
• Biomedical Engineering

Background:

• Human verbal communication relies on complex brain-muscle coordination.
• Individuals with speech loss due to congenital conditions or illness face significant challenges.
• Current speech restoration research using brain-computer interfaces (BCI) is in its nascent stages.

Purpose of the Study:

• To investigate the relationship between acoustic signal chaos parameters and electroencephalography (EEG) signals.
• To identify potential chaos parameters for BCI-based speech restoration.
• To analyze vowel sounds for their correlation with neural signals.

Main Methods:

• Explored regression between chaos parameters of acoustic signals and EEG signals for various vowels.
• Categorized vowels into soft vowels and diphthongs.
• Selected EEG channels with high contribution to the first principal component for analysis.
• Evaluated goodness of fit parameters for regression analysis.

Main Results:

• Identified specific chaos parameters in acoustic and EEG signals corresponding to different vowel sounds.
• Established a regression model to correlate acoustic and neural signal characteristics.
• Determined the suitability of various chaos parameters for BCI applications.

Conclusions:

• The study provides foundational insights into correlating acoustic features with neural signals for speech restoration.
• Findings suggest potential for developing more effective BCI systems for individuals with speech impairments.
• Further research is needed to refine the identified chaos parameters and improve BCI accuracy.