[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Influence of amplitude on place pitch in cochlear implants (CIs)
I have experience roving level with CI subjects. I've tried profile analysis (spectral shape discrimination) with both-back end (current level rove on the electrodes) and front-end (acoustic rove through the sound processor). In the back-end case, it was an profile of current levels on multiple electrodes using CIS. For acoustic profiles, it was the standard Dave Green inspired tone complexes. Die hard profile analysis folks will tell you that it's not REALLY profile analysis if you can't control for level i.e. level rove the stimuli within a trial and subjects are able to hear a spectral change in a single component that's LESS THAN the amount of the rove (see Green "Profile Analysis 1988).
Getting to the pitch thing, roving level in CI patients WITH THE SAME SPECTRAL STIMULI, sends pitch all over creation. Place of stimulation is constant, only level changes, whether done at the front or back end. As a result, CI users can't do "true" profile analysis at all (with one exception out of 9, who was a violin major). This is a bit of a problem in music. Take Bolero for example at a loud level and a soft level. It could sound COMPLETELY DIFFERENT to a CI user, depending on the level.
If you do a complex pitch study and rove level too much, it will be hard to get a systematic result, because every person will respond differently to the level changes. If we could solve the problem-- and keep pitch constant with level changes for CI users, certainly I would expect it to help them appreciate dynamic changes in music. It is, I believe possible to turn off the AGC. I think that will help but not solve the problem.
We do an acoustic test of pitch direction discrimination with complex tones. The level rove is +/- 4 dB. CI users seem to be pretty good at this task. Many can discriminate a semitone direction change. With larger roves, I'm sure they would not do as well.
Nimmons GL, Kang RS, Drennan WR, Longnion J, Ruffin C, Worman T, Yueh B, Rubinstein JT.
Clinical Assessment of Music Perception in Cochlear Implant Listeners.
Otol Neurotol. 2007 Aug 2;Publish Ahead of Print [Epub ahead of print]
On Fri, 7 Dec 2007, Matthias Milczynski wrote:
Dear list,
I am testing pitch ranking in CI subjects. In general, I work with a research
processor and I stream pre-processed stimuli directly to a subject's implant.
To get rid of loudness cues I want to apply amplitude roving. However, when
applying the roving at the front-end (i.e. manipulating the amplitude of
wav-files to be processed by a particular strategy) I observe the following
problem:
Let's assume a complex tone (e.g. piano) at a frequency of 164.8 Hz (E3), with
a duration of 500 ms. Thereby the rms of the sound is relative small (e.g. -30
dB re full scale). Then, let's assume a tone at 130.8 Hz (C3) from the same
instrument and at the same duration but at a higher rms (e.g. 10 dB higher in
rms than the first tone). Due to the non-linear loudness growth function
implemented in these days' CIs (assuming e.g. the ACE strategy with default
adjustments) the stimulation pattern corresponding to the first tone will show
activity on a few apical electrodes (up to about 900 Hz). However the
stimulation pattern corresponding to the second tone (which is higher in
amplitude but lower in pitch) will also show activity at electrodes
corresponding to much higher frequencies (up to approx. 1700 Hz). As a
consequence a subject that mainly relies on place pitch cues could falsely rank
the second tone as the higher one in pitch. That means that amplitude roving at
the front-end can introduce a misleading place pitch cue. Of course the
stimulation patterns for E3 and C3 will differ in temporal pitch cues (i.e. the
frequency of the envelope fluctuations in the E3-pattern will be higher than
for the C3-pattern), however the effectivity of this cue will be
subject-dependent.
Another possibility to amplitude-rove the stimuli would be to apply the roving
at the back-end, i.e. manipulating the electrode current (i.e. multiplication
by a scaling factor) but this type of roving does not necessarily correspond to
a real-life situation and is difficult to implement when working with the
subject's own device.
I would very much appreciate any comments and/or suggestions.
Many thanks!!
Matthias
--------------------------------------------------------------
Matthias Milczynski
PhD Student
ExpORL, Dept. Neurosciences, K.U.Leuven
O.& N2, Herestraat 49 bus 721
B-3000 Leuven
Belgium
Tel:
+32 16 330476
E-mail:
Matthias.Milczynski@xxxxxxxxxxxxxxx
http://gilbert.med.kuleuven.be/Milczynski
--------------------------------------------------------------
Ward R. Drennan, Ph. D.
VM Bloedel Hearing Research Center
Department of Otolaryngology
University of Washington Box 357923
Seattle, WA 98195
Office: (206) 897-1848
Fax: (206) 616-1828