Re: High-frequency hearing in humans (Brian Gygi )


Subject: Re: High-frequency hearing in humans
From:    Brian Gygi  <bgygi@xxxxxxxx>
Date:    Wed, 2 Feb 2011 19:56:16 +0000
List-Archive:<http://lists.mcgill.ca/scripts/wa.exe?LIST=AUDITORY>

----=_vm_0011_W7516821148_1812_1296676576 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Ola Piotr, For environmental sounds, I refer to my 2004 JASA paper (I can send you a= copy if you like) which showed the effect of high-, low- and bandpass fi= ltering on a variety of environmental sounds. I did not test a mosquito b= uzz, but there were several sounds which were highly identifiable even with a highpass filter= cutoff of 8 kHz, such as sirens, rain and coughing. I suspect the siren = and coughing would still be identifiable even if masked by a low-frequenc= y noise as you mention. Rain, since it is essentially a steady state band= pass noise, might not be. Gygi, B., Kidd, G.R. & Watson, C.S. (2004) Spectral-temporal factors in t= he identification of environmental soundsJ. Acoust. Soc. Am. Volume 115, = Issue 3, pp. 1252-1265 As for what the "purpose" of high fequency hearing is, although I dislike= "just-so" stories of adaptive characteristics (though they are fun) it d= oes make sense that the human frequency range would correspond to that of= salient sound sources. Since humans actually have a lower overall freque= ncy response than a number of mammals this suggests those sound sources w= ould be ones for which it necessary to have access to both low and high f= requency information (since the envelope in the two regions are not corre= lated). As far as I know, the only class of sound sources for which this = is generally true is vocalizations. In my research I looked at the cross-= band correlations for a number of sounds, and vocalizations generally hav= e quite low cross-band correlations. Hope you are well, lg Brian Gygi, Ph.D. Speech and Hearing Research Veterans Affairs Northern California Health Care System 150 Muir Road Martinez, CA 94553 (925) 372-2000 x5653 -----Original Message----- From: Piotr Majdak [mailto:piotr@xxxxxxxx Sent: Wednesday, February 2, 2011 11:14 AM To: AUDITORY@xxxxxxxx Subject: Re: High-frequency hearing in humans Dear list, thank you all for the many responses. Below I try to sort and summarize t= he information: Reasons why extended (>8 kHz) high-frequency hearing may be important (be= sides sound localization!) : Improved perceived music/speech naturalness: Moore & Tan (2003) showed th= at reducing the frequency limit from 16 to 10 kHz also reduced the percei= ved naturalness of speech and music. Improved speech perception: Stelmachowicz, Lewis, Choi, & Hoover (2007) s= howed that hearing-impaired children could better distinguish /s/ from /z= / when the hearing aids were configured to 10 rather than to 5 kHz. On th= e other hand, Studebaker and Sherbecoe (1991) showed that in normal heari= ng, frequencies above 8 kHz have a rather low importance for understandin= g (1% in their speech-importance function measured for various SNRs). It = seems like the speech perception may be a factor - probably more data are= needed here, particularly on specific syllables... Detection of high-frequency sounds (like mosquitoes, birds, crickets) esp= ecially when masked with a low-frequency noise (like wind, water). I had = a look at an amp. spectrum of a mosquito sound: most energy was below 4 k= Hz, and low-pass-filtering at 8 kHz still resulted in a mosquito sound. I= wonder which environmental sounds with energy above 8 kHz have no or lit= tle energy below 8 kHz - I have no references yet :-( Distance perception, relying on the frequency-dependent attenuation of so= unds. Nice idea - seems like there are no data available (yet). Better segregation of competing co-located talkers (again, an idea only, = no references available yet) I also received some further related responses like: We could need high-frequency hearing because the bandwidth of human speec= h goes beyond 8 kHz, especially for consonants. This sounds reasonable, h= owever, I'm not sure if the hearing bandwidth is automatically related to= the voice-production bandwidth - for some species is seems to be, for th= e most not... Cochlear amplifier enhances high-frequency tones more than low-frequency = tones. I think that this may be one of the reasons why we can have high-f= requency hearing. On the other hand, more recent studies suggest strong c= ochlear amplification also at lower frequencies... In the middle ear, we have 3 ossicles - having only 1 ossicle (like birds= ) would be hard to transfer frequencies beyond 8 kHz without a strong att= enuation. This again may explain why - having some evolutionary pressure = - mammals were able to develop hearing up to 100 kHz. We are interested in extended high-frequency hearing because it may be go= od for detecting and predicting hearing loss - that's interesting! High-frequency hearing allows humans to do amazing things like echo locat= ion (FlashSonar, you must see the videos, e.g. here). So, it's amazing, but the sound localization appears to be the primary re= ason for our frequency limit. This is strongly supported by the correspon= dence of the head size and the frequency limit for various mammals (see t= his link, thanks to Henry Heffner for the figure+caption), especially con= sidering the more extended frequency range for echo-locators, and the str= ongly reduced frequency range for subterranean mammals. Also, it seems like the role of speech perception - maybe as a secondary = factor - is unclear yet. If you know some references about speech percept= ion with frequencies above 8 kHz, I'd really appreciate that. Some tertiary factors like sound detection may have contributed to the de= velopment of our hearing range. Interestingly, hearing the sound probably= automatically means "wanting to know its direction" - and we land at the= sound localization again. If you know some references about the bandwidt= h of environmental sounds, especially the contribution of frequencies abo= ve 8 kHz, I'd appreciate them, too. Thanks, Piotr Piotr Majdak wrote: Dear list, I'm looking for the reasons for the good high-frequency* hearing in human= s. The reasons I have until now are actually the obvious ones: * Pinna localization cues * Interaural level cues (ILD, they actually start to work from around 2 k= Hz) What do you think: if there were no need for the ILD and pinna cues, woul= d there be any other reasons? Thanks, Piotr *) say, above 8 kHz -- Piotr Majdak Psychoacoustics and Experimental Audiology Acoustics Research Institute Austrian Academy of Sciences Wohllebengasse 12-14, 1040 Vienna, Austria Tel.: +43 1 51581-2511 Fax: +43 1 51581-2530 ----=_vm_0011_W7516821148_1812_1296676576 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable <html><DIV><FONT face=3DVerdana size=3D2></FONT>&nbsp;</DIV> <DIV><FONT size=3D2>Ola&nbsp;Piotr,</FONT></DIV> <DIV><FONT size=3D2></FONT>&nbsp;</DIV> <DIV><FONT size=3D2>For environmental sounds, I refer to my 2004 JASA pap= er (I can send you a copy if you like) which showed the effect of high-, = &nbsp;low- and bandpass filtering on a variety of environmental sounds.&n= bsp; I did not test a mosquito buzz, but there were </FONT></DIV> <DIV><FONT size=3D2>several sounds which were highly identifiable even wi= th a highpass filter cutoff of 8 kHz, such as sirens, rain and coughing.&= nbsp; I suspect the siren and coughing would still be identifiable even i= f masked by a low-frequency noise as you mention.&nbsp; Rain, since it is= essentially a steady state bandpass noise, might not be.</FONT></DIV> <DIV><FONT size=3D2></FONT>&nbsp;</DIV> <H1><FONT face=3Darial,helvetica,sans-serif size=3D2>Gygi, B., Kidd, G.R.= &amp; Watson, C.S. (2004) Spectral-temporal factors in the identificatio= n of environmental sounds</FONT></H1> <DIV><FONT face=3Darial,helvetica,sans-serif size=3D2>J. Acoust. Soc. Am.= Volume 115, Issue 3, pp. 1252-1265</FONT></DIV> <DIV>&nbsp;</DIV> <DIV><FONT size=3D2>As for what the "purpose" of high fequency hearing is= , although I dislike "just-so" stories of adaptive characteristics (thoug= h they are fun) it does make sense that the human frequency range would c= orrespond to that of salient sound sources.&nbsp; Since humans actually h= ave a lower overall frequency&nbsp;response than a number of mammals this= suggests those sound sources would be ones for which it necessary to hav= e access to both low and high frequency information (since the envelope i= n the two regions&nbsp;are not correlated).&nbsp; As far as I know, the o= nly class of sound sources for which this is generally true is vocalizati= ons.&nbsp; In my research I looked at the cross-band correlations for a n= umber of sounds, and vocalizations generally have quite&nbsp;low cross-ba= nd correlations.</FONT></DIV> <DIV><FONT size=3D2></FONT>&nbsp;</DIV> <DIV><FONT size=3D2>Hope you are well,</FONT></DIV> <DIV><FONT size=3D2></FONT>&nbsp;</DIV> <DIV><FONT size=3D2>lg</FONT></DIV> <DIV><FONT size=3D2></FONT>&nbsp;</DIV>Brian Gygi, Ph.D. <BR>Speech and H= earing Research <BR>Veterans Affairs Northern California Health Care Syst= em <BR>150 Muir Road <BR>Martinez, CA 94553 <BR>(925) 372-2000 x5653 <DIV><FONT face=3DVerdana color=3D#0000ff size=3D2></FONT>&nbsp;</DIV> <BLOCKQUOTE style=3D"PADDING-LEFT: 5px; MARGIN-LEFT: 5px; BORDER-LEFT: #0= 000ff 2px solid; MARGIN-RIGHT: 0px"><FONT face=3DTahoma size=3D2>-----Ori= ginal Message-----<BR><B>From:</B> Piotr Majdak [mailto:piotr@xxxxxxxx= <BR><B>Sent:</B> Wednesday, February 2, 2011 11:14 AM<BR><B>To:</B> AUDIT= ORY@xxxxxxxx<BR><B>Subject:</B> Re: High-frequency hearing in huma= ns<BR><BR></FONT> <TABLE class=3D"" id=3D201179 height=3D400 width=3D"100%" bgColor=3D#ffff= ff border=3D0 text=3D"#000000"> <TBODY> <TR> <TD class=3D"" vAlign=3Dtop width=3D"100%"> <P style=3D"MARGIN-BOTTOM: 0.5cm">Dear list,<BR><BR>thank you all for the= many responses. Below I try to sort and summarize the information:<BR><B= R>Reasons why extended (&gt;8 kHz) high-frequency hearing may be importan= t (besides sound localization!) :</P> <UL> <LI> <P style=3D"MARGIN-BOTTOM: 0cm">Improved perceived music/speech naturalne= ss: Moore &amp; Tan (2003) showed that reducing the frequency limit from = 16 to 10 kHz also reduced the perceived naturalness of speech and music.<= /P> <LI> <P style=3D"MARGIN-BOTTOM: 0cm">Improved speech perception: Stelmachowicz= , Lewis, Choi, &amp; Hoover (2007) showed that hearing-impaired children = could better distinguish /s/ from /z/ when the hearing aids were configur= ed to 10 rather than to 5 kHz. On the other hand, Studebaker and Sherbeco= e (1991) showed that in normal hearing, frequencies above 8 kHz have a ra= ther low importance for understanding (1% in their speech-importance func= tion measured for various SNRs). It seems like the speech perception may = be a factor - probably more data are needed here, particularly on specifi= c syllables...</P> <LI> <P style=3D"MARGIN-BOTTOM: 0cm">Detection of high-frequency sounds (like = mosquitoes, birds, crickets) especially when masked with a low-frequency = noise (like wind, water). I had a look at an amp. spectrum of a mosquito = sound: most energy was below 4 kHz, and low-pass-filtering at 8 kHz still= resulted in a mosquito sound. I wonder which environmental sounds with e= nergy above 8 kHz have no or little energy below 8 kHz - I have no refere= nces yet :-(</P> <LI> <P style=3D"MARGIN-BOTTOM: 0cm">Distance perception, relying on the frequ= ency-dependent attenuation of sounds. Nice idea - seems like there are no= data available (yet).</P> <LI> <P>Better segregation of competing co-located talkers (again, an idea onl= y, no references available yet) </P></LI></UL> <P><BR>I also received some further related responses like:</P> <UL> <LI> <P style=3D"MARGIN-BOTTOM: 0cm">We could need high-frequency hearing beca= use the bandwidth of human speech goes beyond 8 kHz, especially for conso= nants. This sounds reasonable, however, I'm not sure if the hearing bandw= idth is automatically related to the voice-production bandwidth - for som= e species is seems to be, for the most not...</P> <LI> <P style=3D"MARGIN-BOTTOM: 0cm">Cochlear amplifier enhances high-frequenc= y tones more than low-frequency tones. I think that this may be one of th= e reasons why we can have high-frequency hearing. On the other hand, more= recent studies suggest strong cochlear amplification also at lower frequ= encies...</P> <LI> <P style=3D"MARGIN-BOTTOM: 0cm">In the middle ear, we have 3 ossicles - h= aving only 1 ossicle (like birds) would be hard to transfer frequencies b= eyond 8 kHz without a strong attenuation. This again may explain why - ha= ving some evolutionary pressure - mammals were able to develop hearing up= to 100 kHz.</P> <LI> <P style=3D"MARGIN-BOTTOM: 0cm">We are interested in extended high-freque= ncy hearing because it may be good for detecting and predicting hearing l= oss - that's interesting!</P> <LI> <P>High-frequency hearing allows humans to do amazing things like echo lo= cation (<A href=3D"http://www.worldaccessfortheblind.org/" target=3D_blan= k mce_href=3D"http://www.worldaccessfortheblind.org/">FlashSonar</A>, you= must see the videos, e.g. <A href=3D"http://dsc.discovery.com/videos/is-= it-possible-real-life-bat-man.html" target=3D_blank mce_href=3D"http://ds= c.discovery.com/videos/is-it-possible-real-life-bat-man.html">here</A>).<= /P></LI></UL> <P>So, it's amazing, but the sound localization appears to be the primary= reason for our frequency limit. This is strongly supported by the corres= pondence of the head size and the frequency limit for various mammals (se= e this <A href=3D"http://piotr.majdak.com/temp/heffner.pdf" target=3D_bla= nk mce_href=3D"http://piotr.majdak.com/temp/heffner.pdf">link</A>, thanks= to Henry Heffner for the figure+caption), especially considering the mor= e extended frequency range for echo-locators, and the strongly reduced fr= equency range for subterranean mammals. <BR><BR>Also, it seems like the r= ole of speech perception - maybe as a secondary factor - is unclear yet. = If you know some references about speech perception with frequencies abov= e 8 kHz, I'd really appreciate that.<BR><BR>Some tertiary factors like so= und detection may have contributed to the development of our hearing rang= e. Interestingly, hearing the sound probably automatically means "wanting= to know its direction" - and we land at the sound localization again. If= you know some references about the bandwidth of environmental sounds, es= pecially the contribution of frequencies above 8 kHz, I'd appreciate them= , too.<BR><BR>Thanks, </P>Piotr<BR><BR><BR>Piotr Majdak wrote: <BLOCKQUOTE cite=3Dmid:20110125092815.1DAEA7966@xxxxxxxx t= ype=3D"cite">Dear list, <BR><BR>I'm looking for the reasons for the good = high-frequency* hearing&nbsp; in humans. <BR><BR>The reasons I have until= now are actually the obvious ones: <BR>* Pinna localization cues <BR>* I= nteraural level cues (ILD, they actually start to work from around 2 kHz)= <BR><BR>What do you think: if there were no need for the ILD and pinna c= ues, would there be any other reasons? <BR><BR>Thanks, <BR><BR>Piotr <BR>= *) say, above 8 kHz <BR><BR></BLOCKQUOTE><BR><BR> <DIV class=3Dmoz-signature>-- <BR>Piotr Majdak<BR>Psychoacoustics and Exp= erimental Audiology<BR><A href=3D"http://www.kfs.oeaw.ac.at/" target=3D_b= lank mce_href=3D"http://www.kfs.oeaw.ac.at">Acoustics Research Institute<= /A><BR><A href=3D"http://www.oeaw.ac.at/" target=3D_blank mce_href=3D"htt= p://www.oeaw.ac.at/">Austrian Academy of Sciences</A><BR>Wohllebengasse 1= 2-14, 1040 Vienna, Austria<BR>Tel.: +43 1 51581-2511<BR>Fax: +43 1 51581-= 2530</DIV> <BLOCKQUOTE></BLOCKQUOTE></TD></TR></TBODY> <BLOCKQUOTE></BLOCKQUOTE> <BLOCKQUOTE></BLOCKQUOTE> <BLOCKQUOTE></BLOCKQUOTE> <BLOCKQUOTE></BLOCKQUOTE> <BLOCKQUOTE></BLOCKQUOTE></BLOCKQUOTE></html> ----=_vm_0011_W7516821148_1812_1296676576--


This message came from the mail archive
/home/empire6/dpwe/public_html/postings/2011/
maintained by:
DAn Ellis <dpwe@ee.columbia.edu>
Electrical Engineering Dept., Columbia University