Transcript Document

Implications of rate pitch on musical
perception and auditory object
formation in cochlear implants
Bomjun Kwon, Trevor Perry
Gallaudet University
Dept of Hearing, Speech and Language
Washington DC
The study was supported by NIDCD (R03 DC009061)
Music in cochlear implants
Image source: FDA.gov
z
z
z
z
z
z
z
z
H
H
H
H
H
H
H
H
1.6 93.6 29.6 49.2 92.0 40.0 93.8 23.2
6
2
2
3
3
3
4
4
5
Bold statement regarding CI users
music appreciation
• They can recognize familiar tunes in
carefully prepared conditions.
– Not many voices, not much harmony
– Dominated by rhythmic components
• The process of music listening involves
pattern-matching (highly top-down)
– More effortful than enjoyable
Fundamental problem
• Perception is based only on place pitch and
does not punctuate the pitch accurately
necessary for musical events
Harmonicity is absent
Temporal pitch as an alternative
• Can temporal pitch work?
– Limited to ~ 300 Hz
– Not as salient as place+temporal (Burns &
Viemeister, 1983)
– Maybe can be used to represent low pitch notes?
– At least the pitch can be accurately represented
– Less top-down pattern matching task
Demonstration of temporal pitch
• Pijl & Schwarz (JASA, 1995; HR, 1995) Pijl (EH
1997)
– CI users can recognize melodies from temporal
pitch sequences presented on single channel
– Their musical interval judgment can be as good as
NH.
• But…. So what?
Temporal harmonics
(Rate harmonics)
• Idea: Deliver rate pitch melody on multiple
channels with harmonic ally related rates
EL17
260.6 pps
293.6 pps
329.6 pps
130.8 pps
Why are we trying this?
To
facilitate
grouping
z
z
z
z
H
H
H
H
8
8
8
8
.
.
.
.
0
6
4
0
on two channels
13
14stimulation
16
13
EL20
130.8 pps
146.8 pps
164.8 pps
130.8 pps
Experiment 1: Conditions
H: Stimulation on two channels
Unison
are more easily grouped if their
Minor 2nd
rates are harmonically related
Major 3rd
Perf 5th
Major 7th
Perf 8th
Procedure
• 4 Nucleus users with substantial musical
experience
• Happy Birthday
• 5 familiar tunes
• Frère Jacques
• All temporal (rate) pitch
• Mary had a little
lamb
• lower note on apical El
• Do-re-mi song
• Higher note on basal El
• When the saints go
marching in
Procedure
• For each melody, 6 buttons are given on
screen, corresponding to 6 conditions
• 6 were randomly arranged in each time.
• Subjects click the button to play
• Repeats are allowed
• Once done, they type consonance rating for
each button—1: complete cacophony, 10:
perfect harmony (almost one sound)
• Each melody was repeated 3 times.
Results
Results
Octave
(harmonic)
Hypothesis supported!
Punchline
• Electrical stimulation on two channels can be
perceptually grouped when stimulation rates
are harmonically related.
Conclusion
• Findings from Experiment 2 are even cooler
•but
Rate
pitch
hastime
potentially
don’t
have
today. important
in auditory
with
CIs.
–implications
Please stay tuned
until nextgrouping
time I give
a talk.
Temporal harmonics… maybe future
Observations / more stories
• Rate pitch percept is not immediate.
– Takes a little “warm-up” time.
• Different range of rates for each subj
– Some prefer higher rate (>200 pps)
• Rates 400~500 pps do enrich sound quality when
used as a harmonic rate even though they alone
cannot elicit pitch percept.
• In terms of sound quality, rates and electrode
position do not appear to be independent.
– They definitely prefer some choice than others.