Checkerboard and interrupted speech: Intelligibility contrasts related to factor-analysis-based frequency bandsa)

Area of Science:

• Auditory Neuroscience
• Speech Perception Research
• Signal Processing in Acoustics

Background:

• Checkerboard speech stimuli involve periodic interruptions in time and frequency.
• Previous studies showed intelligibility varies with frequency bands (2-20) and segment duration (20-320 ms).
• The impact of specific frequency band numbers (4-20) and division parameters on intelligibility remained unclear.

Purpose of the Study:

• To investigate the effects of frequency band numbers and division parameters on checkerboard speech intelligibility.
• To identify the optimal number of frequency bands for speech perception.
• To explore the underlying mechanisms of speech intelligibility using a probability summation model.

Main Methods:

• Systematic evaluation of speech intelligibility using checkerboard stimuli with varying numbers of frequency bands (4-20) and segment durations (20-320 ms).
• Comparison of intelligibility across different frequency division parameters.
• Analysis of intelligibility data using statistical methods and a probability summation model.

Main Results:

• Speech intelligibility was lowest for four-band checkerboard stimuli, except at 320 ms segment duration.
• Temporally interrupted speech and eight-band checkerboard stimuli showed higher intelligibility.
• U-shaped intelligibility curves were observed for four-band and possibly eight-band stimuli.
• Frequency division parameters caused significant intelligibility differences in four-band stimuli at 160 and 320 ms.

Conclusions:

• Factor-analysis-based four frequency bands act as essential speech cue channels for perception.
• A probability summation model may explain the observed U-shaped intelligibility curves.
• Understanding frequency band importance is crucial for developing effective speech processing strategies.