Release from speech-on-speech masking by adding a delayed masker at a different location

Area of Science:

• Auditory Perception
• Acoustics
• Speech Processing

Background:

• Speech masking occurs when competing sounds interfere with auditory comprehension.
• Acoustical reflections, common in real-world environments, can influence speech intelligibility.
• Previous research suggests spatial separation can aid speech perception in noise.

Purpose of the Study:

• To investigate the phenomenon of masking release through simulated acoustical reflections.
• To determine the impact of masker number, signal-to-noise ratio, and delay timing on masking release.
• To explore energetic masking release with speech-spectrum noise as a masker.

Main Methods:

• Three experiments were conducted varying masker presentation (dual, delayed, spatially separated).
• Experiment 1 manipulated the number of masking talkers (1-3) and signal-to-noise (S/N) ratios (-12 to +4 dB).
• Experiments 2 and 3 systematically varied lead-lag masker presentation delay times, using both speech and noise maskers.

Main Results:

• Significant masking release was observed across all tested conditions, with up to 10 dB release for multiple maskers at negative S/N ratios.
• Masking release was substantial for shorter delay times (simulating early reflections) but negligible for longer delays.
• Energetic masking release of up to 4 dB was found for delays of 2 ms or less, diminishing with longer delays.

Conclusions:

• Simulating acoustical reflections via dual, delayed, and spatially separated maskers is an effective strategy for reducing speech masking.
• The timing of simulated reflections is critical, with shorter delays yielding greater masking release.
• This phenomenon has implications for understanding auditory scene analysis and developing assistive listening technologies.