Re: AUDITORY Digest - 11 Jul 2002 to 12 Jul 2002 (#2002-107) (Jont Allen )


Subject: Re: AUDITORY Digest - 11 Jul 2002 to 12 Jul 2002 (#2002-107)
From:    Jont Allen  <jba(at)auditorymodels.org>
Date:    Sat, 13 Jul 2002 01:35:34 -0400

Dear Lars, Masking is composed of two components, and in some sense, they are very dissimilar. First there is the cochlear compression component, which is not really masking in the sense you are speaking of. It is more a cochlear gain factor, that has to do with the "agc" dynamic range control within the cochlea. This controls the level into the neurons. The second is the true masking due to "neural noise." This is the true masking effect, where two signals cannot be separated, because the distributions of detection overlap. The hearing impaired lose the first component, but everyone has the second, as long as they have a brain. ;-) Nice question. --Jont Allen Refs: Here are some references that discuss this in detail, and I would be happy to send links to the eprints of these articles, to anyone who asks: author = {Allen, J.B.} ,title={Psychoacoustics} ,booktitle = {Wiley Encyclopedia of Electrical and Electronics Engineering} ,volume = 17 ,editor={Webster, J.G.} ,publisher={John Wiley \& Sons, Inc} ,address={New York, NY} ,pages = {422-437} ,year = {1999} ,author={Allen, J.B.} ,title={Amplitude compression in hearing aids} ,booktitle={MIT Encyclopedia of Communication Disorders} ,publisher={MIT Press} ,year=2002 ,editor={ R.~Kent } ,chapter={ To appear } ,pages={} ,address={MIT, Boston Ma} author={Allen, J.B.} ,title={Nonlinear Cochlear Signal Processing} ,booktitle={ Physiology of the Ear, Second Edition } ,publisher={Singular Thomson Learning} ,year=2001 ,editor={ Jahn, A.F. and Santos-Sacchi, J.} ,chapter={19} ,pages={393--442} ,address={ 401 West A Street, Suite 325 San Diego, CA 92101 } >Date: Fri, 12 Jul 2002 12:53:17 +0200 >From: =?iso-8859-1?Q?Lars_Bramsl=F8w?= <LAB(at)OTICON.DK> >Subject: Bit-rate reduced audio and hearing loss > >Dear list, > >The modern audio compression techniques (MPEG, MP3, WMF etc..) can = >easily >provide transparent audio quality if the bit-rate is sufficiently high, = >e.g. >256 kbit/s. They are based on more or less advanced hearing models and >exploit the temporal and spectral masking properties of the ear to keep = >all >quantization errors below the masked thresholds. > >If these sounds are used with hearing impaired subjects and hearing = >aids, >one could speculate that either the nature of the hearing loss or the >hearing aid signal processing could potentially unmask the artefacts of = >the >compression. One such example would be a ski-slope hearing loss with = >normal >LF hearing and severe HF hearing loss, combined with the appropriate >frequency shaping, leading to audible artefacts in the normal hearing >region. > >On the other hand, we can expect that the spectral and temporal = >resolution >of the impaired ear is poorer than that of the normal ear. So if the = >audio >quality is transparent to the normal-hearing listener it will also be >perceived as transparent by the hearing-impaired listener. > >Does anyone have good or bad experiences with hearing aid users and = >bit-rate >reduced audio? > >Regards, > >Lars Bramsl=F8w > >----------------------------------------------------- >Lars Bramsl=F8w >Ph.D., M.Sc.E.E. >Audiology > >Oticon A/S >Strandvejen 58 >DK - 2900 Hellerup > >phone: +45 39 13 85 42 >fax: +45 39 27 79 00 > >mailto:lab(at)oticon.dk >http://www.oticon.com >----------------------------------------------------- > >------------------------------ > >Date: Fri, 12 Jul 2002 13:33:44 +0200 >From: alexander lerch <lerch(at)ZPLANE.DE> >Subject: Re: Bit-rate reduced audio and hearing loss > >Hello, > >there was a MP3-Listening Test of the german c't-magazine >published in issue 6/2000 >(http://www.heise.de/ct/00/06/092/default.shtml, in german). > >The listener with the "best" differentiation between original >and coded signals turned out to be hearing impaired >(siginificant hearing loss above 8kHz). > >However, I would not claim this test to be scientific relevant. > >Regards, >Alexander > >Lars Bramsl°w schrieb: > >>Dear list, >> >>The modern audio compression techniques (MPEG, MP3, WMF etc..) can easily >>provide transparent audio quality if the bit-rate is sufficiently high, e.g. >>256 kbit/s. They are based on more or less advanced hearing models and >>exploit the temporal and spectral masking properties of the ear to keep all >>quantization errors below the masked thresholds. >> >>If these sounds are used with hearing impaired subjects and hearing aids, >>one could speculate that either the nature of the hearing loss or the >>hearing aid signal processing could potentially unmask the artefacts of the >>compression. One such example would be a ski-slope hearing loss with normal >>LF hearing and severe HF hearing loss, combined with the appropriate >>frequency shaping, leading to audible artefacts in the normal hearing >>region. >> >>On the other hand, we can expect that the spectral and temporal resolution >>of the impaired ear is poorer than that of the normal ear. So if the audio >>quality is transparent to the normal-hearing listener it will also be >>perceived as transparent by the hearing-impaired listener. >> >>Does anyone have good or bad experiences with hearing aid users and bit-rate >>reduced audio? >> >>Regards, >> >>Lars Bramsl°w >> > > >-- >dipl. ing. >alexander lerch > >zplane.development >http://www.zplane.de >holsteinische str. 39-42 >D-12161 berlin > fon: +49.30.854 09 15.0 > fax: +49.30.854 09 15.5 > >------------------------------ >


This message came from the mail archive
http://www.auditory.org/postings/2002/
maintained by:
DAn Ellis <dpwe@ee.columbia.edu>
Electrical Engineering Dept., Columbia University