Effect of masker temporal pattern, spectrum, and presentation level on speech identification

Area of Science:

• Auditory Perception
• Acoustic Signal Processing
• Psychoacoustics

Background:

• Understanding speech masking is crucial for hearing aid development and auditory scene analysis.
• Masking effects are influenced by the spectral and temporal characteristics of background noise.
• Overall sound level can modulate auditory system performance, affecting masking patterns.

Purpose of the Study:

• To investigate how spectral and temporal dips in maskers affect speech identification.
• To determine the influence of overall sound level on speech masking in different noise types.
• To compare the masking effectiveness of speech-shaped noise (SSN), harmonic complex tones (HCT), and iterated ripple noise (IRN).

Main Methods:

• Sentences were masked by SSN, HCT, and IRN at 50 and 80 dB SPL.
• Signal-to-noise ratios (SNRs) ranged from 0 to -15 dB.
• Speech identification performance was measured to calculate SNR-50 (SNR for 50% correct).

Main Results:

• Speech identification was significantly better with HCT maskers compared to SSN and IRN, especially at SNRs of -5 and -10 dB.
• Higher overall sound levels (80 dB SPL) worsened speech masking for SSN, indicating reduced frequency selectivity.
• At higher levels, HCT maskers showed improved speech intelligibility, suggesting enhanced 'listening in the dips' ability.

Conclusions:

• Spectral and temporal dips in maskers, particularly in HCTs, can significantly improve speech intelligibility.
• The effectiveness of masker dips and the impact of overall sound level on speech masking are interdependent.
• Auditory perception of masked speech is modulated by masker complexity and presentation level.