sound source segregation (determination) the process by which acoustic components are identified as...
Post on 22-Dec-2015
217 views
TRANSCRIPT
![Page 1: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/1.jpg)
Sound source segregation (determination)
The process by which acoustic components are identified as coming
from one or more sound sources.
![Page 2: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/2.jpg)
ANF basal
ANF apical
Red+blue=purple
Time
![Page 3: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/3.jpg)
Two types of method for assessing sound source segregation
• Auditory steaming• Thresholds
![Page 4: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/4.jpg)
Auditory streaming
When two frequencies are close together, you hear them as one source,
or “stream”; when they are far apart you hear two.
![Page 5: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/5.jpg)
A single sound source is perceived
![Page 6: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/6.jpg)
Two sound sources are perceived
![Page 7: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/7.jpg)
Disadvantages of auditory streaming method
• Frequency separation isn’t the only thing that we use to separate sources; this is limited way to study the problem.
• “What does this sound like to you?” doesn’t seem like the sort of question that would produce a reliable answer.
![Page 8: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/8.jpg)
Auditory streaming is a method for studying
• Sound source segregation• Localization• Lateralization• Frequency discrimination
![Page 9: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/9.jpg)
A disadvantage of the auditory streaming method is that
• only tones can be studied• it uses a very subtle perception• it does not produce very reliable results• it cannot be used to study infants or children
![Page 10: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/10.jpg)
If one sound can be segregated from another, then the threshold for the
sound should be lower.
![Page 11: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/11.jpg)
Acoustic cues that could be used to segregate components into sources
• Spectral separation• Temporal separation• Temporal onsets and offsets• Spectral profile• Harmonicity• Spatial separation• Temporal modulations
![Page 12: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/12.jpg)
Spectral profile
From Yost (1994)
![Page 13: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/13.jpg)
Harmonicity
100 200 300
Frequency (Hz)
Am
plit
ude
(dB
)
125 500
Frequency (Hz)
Am
plit
ude
(dB
)250? ?
Frequency (Hz)
Am
plit
ude
(dB
)
![Page 14: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/14.jpg)
Harmonicity
Is this component there?
Is this component there?
Lower threshold
![Page 15: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/15.jpg)
Spatial separation
Component 1Component 2
Component 3
Source 1Source 2
![Page 16: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/16.jpg)
Spatial separation: Masking level difference
Threshold for tone: 25 dB SPL Threshold for tone: 16 dB SPL
!
![Page 17: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/17.jpg)
Spatial separation: Masking level difference
![Page 18: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/18.jpg)
Masking level difference
The MLD is the improvement in audibility that results from dichotic listening
N= noise, S = signal
Monotic = one ear (m)
Diotic = 2 ears, same sound in both (0)
Dichotic = 2 ears, different sound in each (π)
Modified from Gelfand (1998)
![Page 19: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/19.jpg)
Temporal modulations
??
Modified from Yost (1994)
![Page 20: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/20.jpg)
A noise band has a distinct amplitude envelope that we can
isolate and apply to different carrier frequencies
From Yost (1994)
![Page 21: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/21.jpg)
Temporal modulations: Comodulation Masking Release
S = signal, a toneTB = target band, a noise bandCB = cue band, another noise band
Amplitude
Amplitude
If CB is more than a critical band(auditory filter) away from TB infrequency, will the threshold for Schange when CB is present?
From Yost (1994)
![Page 22: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/22.jpg)
Temporal modulations: Comodulation Masking Release
Amplitude
Amplitude
If CB is more than a critical band(auditory filter) away from TB infrequency, will the threshold for Schange when CB is present?
IT DEPENDS.
From Yost (1994)
![Page 23: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/23.jpg)
The target band and the cue band could have the same or different
amplitude envelopes
Different Same = comodulated
From Yost (1994)
![Page 24: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/24.jpg)
Comodulation of noise bands makes threshold for signal lower
Comodulation masking release is the improvement in audibility that results from common amplitude modulation across auditory filters.
From Yost (1994)
![Page 25: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/25.jpg)
The assumption underlying the use of thresholds to study sound
source segregation is
• thresholds are worse when masker and probe can be segregated
• thresholds are better when masker and probe can be segregated
• thresholds vary depending on how the listener listens.
![Page 26: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/26.jpg)
The masking level differences demonstrates the importance of which of the following for sound
source segregation?• Temporal onsets• Temporal modulation• Spectral profile• Spatial location
![Page 27: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/27.jpg)
Which of these demonstrates the importance of temporal onsets to
sound source segregation?
• Masking level difference• Comodulation masking release• Forward fringe masking• Profile analysis
![Page 28: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/28.jpg)
Comodulation masking release demonstrates the importance of which of the following for sound
source segregation?
• Temporal onsets• Temporal modulation• Spectral profile• Spatial location
![Page 29: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/29.jpg)
How does the brain identify features of sound that allow us to segregate them?• Spatial separation – calculation of IID, ITD in
MSO and LSO• Temporal separation• Temporal onsets and offsets• Temporal modulations• Spectral profile• Harmonicity• Spectral separation
![Page 30: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/30.jpg)
The cochlear nucleus
From Pickles (1988)
![Page 31: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/31.jpg)
Cell types in the cochlear nucleus
From Pickles (1988)
![Page 32: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/32.jpg)
Cell types in the
cochlear nucleus
From Cant (1992)
![Page 33: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/33.jpg)
![Page 34: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/34.jpg)
Conclusions• The auditory system analyzes sounds into
their component frequencies and then segregates the components that belong together.
• Sound source segregation affects our ability to detect sound.
• Sound source segregation depends on the ability of the system to identify the spectrum, spectral shape (profile), amplitude envelope, pitch and location of sounds.
• The extraction of these features begins at the lowest level of the auditory pathways.
![Page 35: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/35.jpg)
Some facts about sound source segregation relevant to cochlear
implantsWill people with cochlear implants
have trouble segregating sound sources?
![Page 36: Sound source segregation (determination) The process by which acoustic components are identified as coming from one or more sound sources](https://reader030.vdocuments.us/reader030/viewer/2022032523/56649d785503460f94a5b751/html5/thumbnails/36.jpg)
It depends on whether they have the code for these cues
• Temporal separation• Temporal onsets and offsets• Temporal modulations• Spectral profile• Spectral separation• Harmonicity• Spatial separation