Related Experiment Videos
Frequency-weighting functions for broadband speech as estimated by a correlational method
C W Turner1, B J Kwon, C Tanaka
1Department of Speech Pathology and Audiology, University of Iowa, Iowa City 52241, USA.
The Journal of the Acoustical Society of America
|September 24, 1998
Summary
Listeners utilize specific frequency bands more heavily for speech recognition. Normal-hearing individuals demonstrate consistent frequency weighting strategies when processing broadband speech signals.
Area of Science:
- Auditory Neuroscience
- Speech Perception
- Psychoacoustics
Background:
- Speech recognition relies on processing information across the entire frequency spectrum.
- Understanding how listeners weight different frequency bands is crucial for auditory models.
Purpose of the Study:
- To determine the relative contributions of different frequency bands to speech recognition.
- To investigate listener strategies for combining auditory information across frequency regions.
Main Methods:
- Utilized a correlational method with the Nonsense Syllable Test materials.
- Filtered speech into four frequency bands, adding noise to each band independently.
- Calculated point biserial correlations between signal-to-noise ratio and listener responses for each band.
Main Results:
- Individual frequency bands contributed equally when presented alone.
- Significant differences in band contributions emerged when bands were combined (broadband).
- Listeners showed a strong reliance on specific frequency bands in the broadband condition.
Conclusions:
- Normal-hearing listeners employ distinct frequency-weighting strategies for speech recognition.
- These strategies involve selective attention and integration of auditory information across frequency.
- Frequency weighting functions were remarkably consistent across participants.