response! (Hamish Innes-Brown )


Subject: response!
From:    Hamish Innes-Brown  <hinnesbrown(at)bsi.swin.edu.au>
Date:    Mon, 2 Sep 2002 14:25:14 +1000

This is a multi-part message in MIME format. ------=_NextPart_000_002F_01C2528C.8DCF7E40 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Thanks very much for all the responses! There was quite alot of useful information sent to me as well as the list (actually, a torrent might be a better description), so I thought I might summarise and send to the list in an attempt to "give something back"... Hope this is useful to someone and not repetition - I'm learning rapidly from a fairly uninformed position! The comments fell into three main categories: Firstly I had several people suggest (sensibly) that I just avoid the whole problem and use speakers. I forgot to mention in my initial email that it's imperitive that headphones be used as there will be space constraints, otherwise this is exactly what I would do! Secondly it was pointed out several times that it was the mismatch between a "standard" HRTF and the real HRTF for an individuial that might result in poor externalisation. I downloaded the CIPIC (Center for Image Processing and Integrated Computing at the University of California) database which contains HRTF data for 43 human subjects and two KEMAR dummies (large/small pinnae). I did indeed find that (just me listening with headphones, no flashing lights) some subjects' data worked better for me than others, with the sense of "externalisation" or distance being the main thing that varied (although the left/right alignment seemed to vary quite a bit as well). However, as was pointed out, having the synchronised flashing lights would probably help in this reguard (ventriloquist effect). Thirdly - adding uncorrelated (left/right) reverberation to "standard" HRTF filtered sound may help avoid having to measure HRTF for individuals, yet possibly help to externalise sounds with only a minum effort. One person pointed out the Simon Effect, where responses are speeded on the ipsilateral side of stimulation. This is a good point: the Simon Effect can usually be defeated by appropriately mixing/randomising the design within subjects. Another pointed out the impoprtance of preventing head movements, to reduce the possibility of divergence between the auditory and visual display when headphones are used. SO given that I need only an approximate solution for this application, what I'm going to investigate is the following: I'll use the CIPIC database (probably one of the KEMAR dummy sets) and investigate the judicious application of uncorrelated reverb, remembering that I have the advatage of having synchronised flashing lights to assist with the front/back confusions that might normally result from using only an approximate HRTF filter. Hopefully that will give me enough of a sense of the sound being outside the head while still using headphones. ANd of course the head will be restrained (also eye-movements monitored but thats another story), and the design appropriately mixed :) thanks once again for all the comments -hamish- ------=_NextPart_000_002F_01C2528C.8DCF7E40 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN"> <HTML><HEAD> <META http-equiv=3DContent-Type content=3D"text/html; = charset=3Diso-8859-1"> <META content=3D"MSHTML 6.00.2719.2200" name=3DGENERATOR></HEAD> <BODY> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console" = size=3D2>Thanks=20 very much for all the responses!</FONT></SPAN></DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console"=20 size=3D2></FONT></SPAN>&nbsp;</DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console" = size=3D2>There was=20 quite alot of useful information&nbsp;sent to me as well as the list = (actually,=20 a torrent might be a better description),&nbsp;so I thought I might = summarise=20 and send to the list in an attempt to "give something back"...&nbsp; = Hope this=20 is useful to someone and not repetition - I'm learning rapidly from a = fairly=20 uninformed position!</FONT></SPAN></DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console"=20 size=3D2></FONT></SPAN>&nbsp;</DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console" = size=3D2>The=20 comments fell into three main categories:</FONT></SPAN></DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console"=20 size=3D2></FONT></SPAN>&nbsp;</DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console"=20 size=3D2></FONT></SPAN>&nbsp;</DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console"=20 size=3D2></FONT></SPAN>&nbsp;</DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console" = size=3D2>Firstly I=20 had several people suggest (sensibly) that I just avoid the whole = problem and=20 use speakers.&nbsp; I forgot to mention in my initial email that it's=20 imperitive&nbsp;that headphones be used as there will be space = constraints,=20 otherwise this is exactly what I would do!</FONT></SPAN></DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console"=20 size=3D2></FONT></SPAN>&nbsp;</DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console" = size=3D2>Secondly=20 it&nbsp;was pointed out several times&nbsp;that it was the mismatch = between a=20 "standard" HRTF and the real HRTF for an individuial that might result = in poor=20 externalisation.&nbsp; I downloaded the CIPIC (Center for Image = Processing and=20 Integrated Computing at the University of California) database which = contains=20 HRTF data for 43 human subjects and two KEMAR dummies (large/small=20 pinnae).&nbsp; I did indeed find that (just me listening with = headphones, no=20 flashing lights) some subjects' data worked better for me than others, = with the=20 sense of "externalisation" or distance&nbsp;being the main thing that = varied=20 (although the left/right alignment seemed to vary quite a bit as well). = However,=20 as was pointed out, having the synchronised flashing lights would = probably help=20 in this reguard (ventriloquist effect).</FONT></SPAN></DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console"=20 size=3D2></FONT></SPAN>&nbsp;</DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console" = size=3D2>Thirdly -=20 adding uncorrelated (left/right) reverberation to "standard" HRTF = filtered sound=20 may help avoid having to measure HRTF for individuals, yet possibly help = to=20 externalise sounds with only a minum effort.</FONT></SPAN></DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console"=20 size=3D2></FONT></SPAN>&nbsp;</DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console" = size=3D2>One=20 person pointed out the Simon Effect, where responses are speeded on the=20 ipsilateral side of stimulation.&nbsp; This is a good point: the Simon = Effect=20 can usually be defeated by appropriately mixing/randomising the design = within=20 subjects.</FONT></SPAN></DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console"=20 size=3D2></FONT></SPAN>&nbsp;</DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console" = size=3D2>Another=20 pointed out the impoprtance of preventing head movements, to reduce the=20 possibility of divergence between the auditory and visual display when=20 headphones are used.</FONT></SPAN></DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console"=20 size=3D2></FONT></SPAN>&nbsp;</DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console"=20 size=3D2></FONT></SPAN>&nbsp;</DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console"=20 size=3D2></FONT></SPAN>&nbsp;</DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console"=20 size=3D2></FONT></SPAN>&nbsp;</DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console" = size=3D2>SO given=20 that I need only an approximate solution for this application, what I'm = going to=20 investigate is the following:&nbsp; I'll use the CIPIC database = (probably one of=20 the KEMAR dummy sets) and investigate the judicious application of = uncorrelated=20 reverb, remembering that I have the advatage of having synchronised = flashing=20 lights to assist with the front/back confusions that might normally = result from=20 using only an approximate HRTF filter.&nbsp; Hopefully that will give me = enough=20 of a sense of the sound being outside the head while still using=20 headphones.&nbsp; ANd of course the head will be restrained (also = eye-movements=20 monitored but thats another story), and the design appropriately mixed=20 :)</FONT></SPAN></DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console"=20 size=3D2></FONT></SPAN>&nbsp;</DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console"=20 size=3D2></FONT></SPAN>&nbsp;</DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console"=20 size=3D2></FONT></SPAN>&nbsp;</DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console" = size=3D2>thanks=20 once again for all the comments</FONT></SPAN></DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console"=20 size=3D2></FONT></SPAN>&nbsp;</DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console"=20 size=3D2>-hamish-</FONT></SPAN></DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console"=20 size=3D2></FONT></SPAN>&nbsp;</DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console"=20 size=3D2></FONT></SPAN>&nbsp;</DIV> <DIV><SPAN class=3D549580401-02092002></SPAN>&nbsp;</DIV> <DIV><SPAN class=3D549580401-02092002><FONT face=3D"Lucida Console"=20 size=3D2></FONT></SPAN>&nbsp;</DIV></BODY></HTML> ------=_NextPart_000_002F_01C2528C.8DCF7E40--


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