Please enable JavaScript.
Coggle requires JavaScript to display documents.
Frequency selectivity and masking - Coggle Diagram
Frequency selectivity and masking
Critical bands (Auditory Filters)
No actual filters in humans
Still representative of perception
Fletcher's Experiment
Threshold increases with noise bandwidth, up to a certain point - Saturation
Noise centered at pure tone stimulus freq. and with varying bandwidth
Power spectrum model of masking
Patterson and Moore (1986)
P_s = k
N_0
\del f_CB
P_s = Signal power
k = Fixed SNR
N_0 = Noise masker density
\del f_CB = Critical Bandwidth
Assumptions
Peripheral auditory system contains an array of linear bandpass filters.
The listener is assumed to use just one filter with CF tuned to the signal frequency (highest SNR).
Only the components in the noise that pass through the filter have any effect in masking the signal
Masked threshold corresponds to fixed SNR, k, at the output of the filter.
Filter shape and bandwidth
Estimating Critical Bands (ERB)
Using amplitude modulation
Determine whether the sidebands are resolved or masked (whether they fall into the filter band)
Narrow noise on stim freq. Increase BW until threshold does not increase anymore. That is the BW
Loudness comparison
Freqs in filter BW have loudness calculated from total intensity
If not, loudness is loudness summation
Estimating shape
Notched noise
Noise power falls into the band if notch too narrow, thus filter shape is determined
Power spectrum model of masking
Strictly linear
Integrates the power centered at CF
Notched noise masking
RoEX Filter
Rounded exponential
Empirical accuracy (derived from data)
ERB
0.89 mm on the BM
Masking- and excitation patterns
Auditory filters vs excitation patterns
EX. Pats -> broaden towards high freqs. Upwards spread of masking
Hearing impairment and frequency selectivity
Filters are very broad, almost flat outside CF
Very low selectivity
Filter amount diminished, details removed
Overview
Frequency selectivity represents fundamental characteristic of mammalian auditory system.
• Central to our ability to hear one sound in the presence of other
• Key role in many other aspects of auditory perception (speech, music)
• Frequency selectivity detoriated in hearing-impaired listeners.
• Frequency selectivity often demonstrated and measured by studying masking.
• Frequency-selective analysis important for many applications.
Summary / Applications
Objective speech and sound quality
Clinical diagnostics
Automatic speech recognition
Hearing aids
Perceptual coders
Masking
Upwards spread
Mask one sound with another