Related Experiment Videos
Neural system identification model of human sound localization
1Department of Physiology, and Computer and Engineering Laboratory, The University of Sydney, New South Wales, Australia. craigj@physiol.usyd.edu.au
The Journal of the Acoustical Society of America
|September 29, 2000
Summary
Biological constraints shape human auditory localization. A neural model using head-related transfer functions and frequency division achieved humanlike sound localization performance, highlighting the importance of these constraints.
Area of Science:
- Auditory Neuroscience
- Computational Auditory Neuroscience
Background:
- Human sound localization relies on directional acoustical cues.
- Understanding the biological constraints of this process is crucial for accurate modeling.
Purpose of the Study:
- To investigate the role of biological constraints in human auditory localization.
- To compare the performance of competing models against human subjects using biologically plausible realism constraints.
Main Methods:
- Utilized a psychophysical and neural system modeling approach.
- Derived directional acoustical cues from human head-related transfer functions (HRTFs).
- Employed the Auditory Image Model for cochlear processing and a time-delay neural network for spatial analysis.
Main Results:
- A system model combining cochlear processing and neural networks achieved humanlike sound localization.
- Frequency division and training with variable bandwidth/center-frequency sounds were key for model performance.
- Results underscore the relevance of these factors to human sound localization.
Conclusions:
- Biological realism constraints are integral to accurate auditory localization models.
- Frequency division and adaptable training parameters significantly influence sound localization accuracy.
- The developed model provides insights into the neural mechanisms underlying human spatial hearing.