Frequency-importance functions for words in high- and low-context sentences

Area of Science:

• Audiology
• Speech Science
• Psychoacoustics

Background:

• Understanding speech intelligibility is crucial for hearing aid development and auditory rehabilitation.
• The influence of semantic context on auditory perception remains an area of active research.
• Previous studies have explored spectral weighting for speech intelligibility, but context effects require further investigation.

Purpose of the Study:

• To determine the relative importance of different spectral regions for speech intelligibility.
• To compare frequency-importance functions under neutral versus predictable sentence contexts.
• To investigate the impact of semantic predictability on speech perception in noise.

Main Methods:

• Developed frequency-importance functions using Articulation Index (AI) methods.
• Presented monosyllabic words in sentences with varying predictive information (neutral vs. predictable).
• Tested 42 young normal-hearing adults across a range of signal-to-noise ratios (-8 to +14 dB) with speech-shaped noise.

Main Results:

• Semantic context showed only slight differences in 1/3-octave importance functions.
• The crossover frequencies between importance functions shifted by a constant 180 Hz.
• Results suggest context influences frequency-importance relationships and intelligibility dynamic range.

Conclusions:

• Semantic context, under the tested conditions, minimally alters the fundamental frequency-importance functions for speech intelligibility.
• The observed shift in crossover frequencies indicates a subtle but consistent effect of predictability.
• The Articulation Index model's parameter estimation and theoretical underpinnings are discussed in light of these findings.