Different auditory filter bandwidth estimates based on profile analysis, notched noise, and hybrid tasks

Area of Science:

• Psychoacoustics
• Auditory Perception
• Signal Processing

Background:

• Auditory filter bandwidth is crucial for understanding sound perception.
• Previous models have not fully accounted for variations in auditory filter bandwidth across different listening tasks.

Purpose of the Study:

• To estimate auditory filter bandwidths using three distinct experimental paradigms.
• To compare bandwidth estimates derived from profile analysis, notched-noise, and hybrid masking experiments.
• To evaluate the influence of level randomization on auditory filter bandwidth estimation.

Main Methods:

• Conducted three experiments: profile analysis, notched-noise detection, and a hybrid masking task.
• Employed sinusoidal components and noise maskers with varying frequency ranges and spectral characteristics.
• Utilized a variant of Durlach et al.'s channel model with a two-parameter roex(p,r) filter weighting function to estimate bandwidths.

Main Results:

• Auditory filter bandwidth estimates were approximately two to three times larger for the notched-noise and hybrid detection tasks compared to the profile-analysis task.
• Level randomization had only modest effects on detection thresholds in the notched-noise and hybrid experiments.

Conclusions:

• Auditory filter bandwidth is not constant and varies depending on the listening task.
• The findings necessitate adjustments to auditory models to account for task-dependent bandwidth variations.
• The roex(p,r) filter function effectively models auditory filtering across different experimental conditions.