How Anxiety State Influences Speech Parameters: A Network Analysis Study from a Real Stressed Scenario

Area of Science:

• Psychology
• Speech Science
• Computational Linguistics

Background:

• Traditional voice analysis for anxiety assessment yields inconsistent results due to focus on isolated acoustic features.
• A network perspective offers a novel approach to understanding the complex relationships between anxiety states and vocal parameters.

Purpose of the Study:

• To explore the relationship between anxiety states and vocal parameters using network analysis in ecologically valid settings.
• To identify potential speech markers for anxiety assessment by examining vocal parameter interactions.

Main Methods:

• A cross-sectional study involving 316 undergraduate students who performed a picture description task.
• Speech recordings were analyzed for five acoustic parameters: jitter, fundamental frequency (F0), formant frequencies (F1/F2), intensity, and speech rate.
• Network analysis was employed to examine the relationships between these parameters and self-reported anxiety levels.

Main Results:

• A robust negative relationship was found between jitter and state anxiety, with jitter being the sole consistent speech parameter linked to anxiety.
• Higher anxiety levels were associated with a coupled network of intensity and F1/F2, unlike the sparser network in low-anxiety states.
• State anxiety appears to directly influence jitter and restructure speech feature relationships.

Conclusions:

• Speech parameter networks can aid in recognizing anxiety in real-world settings.
• The negative jitter-anxiety relationship and the intensity-F1/F2 coupling in high-anxiety states suggest potential speech-based anxiety markers.
• Examining vocal parameter interactions, particularly jitter, is crucial for effective speech-based anxiety detection.