Re: Is there considerable phase locking up to 6 kHz? ("Richard F. Lyon" )


Subject: Re: Is there considerable phase locking up to 6 kHz?
From:    "Richard F. Lyon"  <DickLyon(at)ACM.ORG>
Date:    Thu, 18 Mar 2004 08:33:15 -0800

--============_-1132494356==_ma============ Content-Type: text/plain; charset="iso-8859-1" ; format="flowed" Content-Transfer-Encoding: quoted-printable At 9:53 AM +0100 03/18/2004, Eckard Blumschein wrote: >However, we all should be wary of thoughtlessly using notions like spectrum >and temporal fine structure. I'm glad that we have that point of agreement. >Already the fundamentally inappropriate traditional spectrogram illustrates >that the iteration of a segment of noise without any spectral profile >introduces an audible spectral signature. Of course, the FCT-based natural >spectrogram shows a more realistic picture of firing pattern in the >auditory nerve. I remind those who do not trust in FCT, because they >wrongly put it in the drawer of an exotic mathematical idea while it >actually replaces FT, of the need to define what we are talking about if we >are using terms like spectral component. I'm trying to decide whether that comment is=20 directed at me. Certainly I've always been very=20 careful about concepts such as spectral=20 components, as I believe the such=20 frequency-domain concepts often force thinking=20 into wrong directions. I also don't regard=20 =46ourier Cosine Transforms as any more exotic than=20 other mathematical transforms. But I do mistrust=20 them as auditory models. >Martin Braun is certainly correct in that, there are at least two main >streams of auditory information within each CN. However, he apparently >ignores tonotopy as long as he doesn't follow my suggestion that place code >is the best base for subsequent temporal processing. For more than a >century, Fourier analysis and place code were considered the basis of >hearing and of related audio technology because alternative temporal models >failed. Here we have a different view of the impact of=20 the Fourier analysis approach and place code on=20 the historical development of auditory theory.=20 I'm more aligned with a quote that I heard=20 attributed to Georg von B=E9k=E9sy: "Dehydrated cats=20 and the application of Fourier analysis to=20 problems in hearing become more and more a=20 handicap for progress in hearing research." As far as I know, temporal models have succeeded=20 more than failed (that is, temporal models of=20 processing in and beyond the cochlea, not to be=20 confused with temporal processing of raw sound=20 waveforms). Spectral models, while widely used,=20 often run into limitations that make them "fail". >Why not seriously dealing with an unseen mathematically correct and >physiologically plausible model that unites function of cochlea and brain >in a somewhat strange hidden manner which is already known as cepstral >analysis? It also may elucidate why different codes contribute to a unitary >pitch. Cepstral analysis is fine as far as it goes. But=20 it is rather limiting, as a mathematical=20 framework that stops short of describing what's=20 going on in detail, essentially ignoring temporal=20 fine structure on the auditory nerve. >I do not appreciate glossing over FCT as a red herring since such emotional >arguments are difficult to falsify. Nonetheless, I would hope that expert >listeners confirm or deny the putative 400/800 Hz confusion. So far, I am >only aware of a plausible 50/100 Hz confusion in case of iterated noise >segments with alternating polarity inversion (Warren & Wrightson 1981). OK, this part I'm certain is directed at me,=20 since I mentioned red herrings. My comment was=20 not directed at the FCT itself, which is a=20 perfectly fine transform, but rather to the idea=20 that if you use it then you can ignore questions=20 of temporal fine structure. Specifically, what I=20 said quoted your assertion, this way ...the idea=20 that "FCT is the only realistic cochlear=20 transform" is just a mathematician trying to=20 force a biological system to be something he can=20 analyze.' I apologize for putting it in such=20 personal terms. Dick >Eckard Blumschein --============_-1132494356==_ma============ Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable <!doctype html public "-//W3C//DTD W3 HTML//EN"> <html><head><style type=3D"text/css"><!-- blockquote, dl, ul, ol, li { padding-top: 0 ; padding-bottom: 0 } --></style><title>Re: Is there considerable phase locking up to 6 kHz?</title></head><body> <div>At 9:53 AM +0100 03/18/2004, Eckard Blumschein wrote:</div> <blockquote type=3D"cite" cite>However, we all should be wary of thoughtlessly using notions like spectrum</blockquote> <blockquote type=3D"cite" cite>and temporal fine structure.</blockquote> <div><br></div> <div><br></div> <div>I'm glad that we have that point of agreement.</div> <div><br> <br> </div> <blockquote type=3D"cite" cite>Already the fundamentally inappropriate traditional spectrogram illustrates<br> that the iteration of a segment of noise without any spectral profile<br> introduces an audible spectral signature. Of course, the FCT-based natural<br> spectrogram shows a more realistic picture of firing pattern in the<br> auditory nerve. I remind those who do not trust in FCT, because they<br> wrongly put it in the drawer of an exotic mathematical idea while it<br> actually replaces FT, of the need to define what we are talking about if we</blockquote> <blockquote type=3D"cite" cite>are using terms like spectral component.</blockquote> <div><br> <br> </div> <div>I'm trying to decide whether that comment is directed at me.&nbsp; Certainly I've always been very careful about concepts such as spectral components, as I believe the such frequency-domain concepts often force thinking into wrong directions.&nbsp; I also don't regard =46ourier Cosine Transforms as any more exotic than other mathematical transforms.&nbsp; But I do mistrust them as auditory models.</div> <div><br></div> <div><br></div> <blockquote type=3D"cite" cite>Martin Braun is certainly correct in that, there are at least two main<br> streams of auditory information within each CN. However, he apparently<br> ignores tonotopy as long as he doesn't follow my suggestion that place code<br> is the best base for subsequent temporal processing. For more than a<br> century, Fourier analysis and place code were considered the basis of<br> hearing and of related audio technology because alternative temporal models<br> failed.</blockquote> <div><br> <br> </div> <div>Here we have a different view of the impact of the Fourier analysis approach and place code on the historical development of auditory theory.&nbsp; I'm more aligned with a quote that I heard attributed to<font face=3D"Arial" size=3D"-3" color=3D"#000000"> Georg von B=E9k=E9sy</font>:&nbsp; &quot;Dehydrated cats and the application of =46ourier analysis to problems in hearing become more and more a handicap for progress in hearing research.&quot;</div> <div><br></div> <div>As far as I know, temporal models have succeeded more than failed (that is, temporal models of processing in and beyond the cochlea, not to be confused with temporal processing of raw sound waveforms).&nbsp; Spectral models, while widely used, often run into limitations that make them &quot;fail&quot;.</div> <div><br></div> <div><br></div> <blockquote type=3D"cite" cite>Why not seriously dealing with an unseen mathematically correct and<br> physiologically plausible model that unites function of cochlea and brain<br> in a somewhat strange hidden manner which is already known as cepstral<br> analysis? It also may elucidate why different codes contribute to a unitary</blockquote> <blockquote type=3D"cite" cite>pitch.</blockquote> <div><br></div> <div><br></div> <div>Cepstral analysis is fine as far as it goes.&nbsp; But it is rather limiting, as a mathematical framework that stops short of describing what's going on in detail, essentially ignoring temporal fine structure on the auditory nerve.</div> <div><br> <br> </div> <blockquote type=3D"cite" cite>I do not appreciate glossing over FCT as a red herring since such emotional<br> arguments are difficult to falsify. Nonetheless, I would hope that expert<br> listeners confirm or deny the putative 400/800 Hz confusion. So far, I am<br> only aware of a plausible 50/100 Hz confusion in case of iterated noise</blockquote> <blockquote type=3D"cite" cite>segments with alternating polarity inversion (Warren &amp; Wrightson 1981).</blockquote> <div><br></div> <div><br></div> <div>OK, this part I'm certain is directed at me, since I mentioned red herrings.&nbsp; My comment was not directed at the FCT itself, which is a perfectly fine transform, but rather to the idea that if you use it then you can ignore questions of temporal fine structure.&nbsp; Specifically, what I said quoted your assertion, this way ...the idea that &quot;FCT is the only realistic cochlear transform&quot; is just a mathematician trying to force a biological system to be something he can analyze.'&nbsp; I apologize for putting it in such personal terms.</div> <div><br></div> <div>Dick</div> <div><br> <br> </div> <blockquote type=3D"cite" cite>Eckard Blumschein</blockquote> <div><br></div> </body> </html> --============_-1132494356==_ma============--


This message came from the mail archive
http://www.auditory.org/postings/2004/
maintained by:
DAn Ellis <dpwe@ee.columbia.edu>
Electrical Engineering Dept., Columbia University