Related Experiment Videos
Masked detection thresholds and temporal integration for noise band signals
C Formby1, M G Heinz, C E Luna
1Department of Otolaryngology-Head & Neck Surgery, Johns Hopkins University School of Medicine, Baltimore, Maryland 21203.
The Journal of the Acoustical Society of America
|July 1, 1994
Summary
Listeners use multiple cues, including timing and spectral shape, to detect sounds in noise, improving performance beyond simple energy detection. This finding aids in understanding auditory perception and developing new detection models.
Area of Science:
- Auditory perception
- Psychoacoustics
- Signal detection theory
Background:
- Masked detection thresholds are crucial for understanding auditory perception in noisy environments.
- Speech acoustics involve signals with varying bandwidths and durations, necessitating studies on their detection.
- Previous models often assume a constant bandwidth-duration product, which may not fully capture human performance.
Purpose of the Study:
- To measure masked detection thresholds for signals with varying bandwidths and durations in noise.
- To investigate the role of masker level variation on detection performance.
- To identify and quantify the contribution of different listening cues in masked detection.
Main Methods:
- Masked detection thresholds were measured using noise signals of different bandwidths (62-6000 Hz) and durations (10-480 ms).
- Signals were presented in uncorrelated maskers with either constant or randomly varying spectrum levels.
- Temporal integration functions were fitted to estimate time constants (tau) and analyze detection cues.
Main Results:
- Detection performance was disrupted in random-level maskers under specific simultaneous gating and spectral matching conditions.
- Estimated time constants (tau) showed an inverse relationship with bandwidth (W) for bandwidths wider than the critical bandwidth.
- Detection did not follow a constant bandwidth-duration (WT) product, but was influenced by relative signal and masker parameters.
- Relative timing and spectral shape cues provided a significant advantage (10-12 dB) over traditional energy cues alone.
Conclusions:
- Listeners utilize multiple auditory cues (timing, spectral shape, energy) for enhanced sound detection in noise.
- A new multi-cue detection model accurately predicts empirical masked detection thresholds (r=0.95).
- These findings advance the understanding of auditory signal processing and inform the development of more sophisticated detection models.