EEG-based Objective Quality Assessment of Interrupted Speech

Area of Science:

• Auditory Neuroscience
• Speech Processing
• Psychoacoustics

Background:

• Speech interruption degrades acoustic information, impacting perceived speech quality.
• Current methods for assessing interrupted speech quality primarily rely on subjective evaluations.
• Objective measures are needed to complement subjective assessments in understanding speech perception under degradation.

Purpose of the Study:

• To objectively investigate the impact of interruption duty cycle and rate on perceived speech quality.
• To explore the potential of event-related potentials (ERPs), specifically mismatch negativity (MMN), as an objective indicator of interrupted speech quality.
• To correlate objective ERP measures with subjective quality ratings.

Main Methods:

• An oddball-paradigm based ERP experiment was conducted with periodically interrupted speech stimuli.
• Participants listened to stimuli under varying interruption duty cycles and rates.
• Subjective quality ratings were collected concurrently with ERP data acquisition.

Main Results:

• Both interruption duty cycle and interruption rate significantly affected the peak latency of the mismatch negativity (MMN) ERP response.
• A smaller interruption duty cycle or a larger interruption rate resulted in a shorter MMN peak latency.
• These objective findings demonstrated strong consistency with the subjective quality ratings provided by participants.

Conclusions:

• The peak latency of the mismatch negativity (MMN) response serves as a reliable objective indicator for the perceived quality of periodically interrupted speech.
• Objective ERP measures can effectively reflect the impact of speech interruption parameters on auditory perception.
• This research provides a foundation for developing objective tools for speech quality assessment in degraded conditions.