Re: AUDITORY Digest - 25 Jan 2014 to 26 Jan 2014 (#2014-29) (Andy Sabin )

Subject: Re: AUDITORY Digest - 25 Jan 2014 to 26 Jan 2014 (#2014-29) From: Andy Sabin <asabin@xxxxxxxx> Date: Mon, 27 Jan 2014 22:53:15 -0600 List-Archive:<http://lists.mcgill.ca/scripts/wa.exe?LIST=AUDITORY> --bcaec548a62757d44904f1009bf3 Content-Type: text/plain; charset=ISO-8859-1 Many thanks to all for super thoughtful responses! For clarification, I was interested in typical SNRs for going out to a restaurant/bar. A few folks have asked for a summary... Seems like there are two main points: (1) there are many sources of variation in the talker/listener/environment that can have an enormous effect on SNR, and (2) there is a principled reason to think that, on average, the SNR often ends up near 0 dB (in situations that are limited by noise) . There might be a fun informal experiment you can try the next time you are at a restaurant/bar: Measure the SPL near your mouth when talking (there are many free apps that are decent), estimate the level of your voice at the listener's ear using the mouth-to-ear distance, and compare that level to the SPL at the listener's ear when you are not talking. If you do that, I'd be interested to know what you find (off list please). On Sun, Jan 26, 2014 at 11:04 PM, AUDITORY automatic digest system < LISTSERV@xxxxxxxx> wrote: > Date: Mon, 27 Jan 2014 00:04:14 -0500 > Reply-To: AUDITORY - Research in Auditory Perception < > AUDITORY@xxxxxxxx> > Sender: AUDITORY - Research in Auditory Perception < > AUDITORY@xxxxxxxx> > From: AUDITORY automatic digest system <LISTSERV@xxxxxxxx> > Subject: AUDITORY Digest - 25 Jan 2014 to 26 Jan 2014 (#2014-29) > To: AUDITORY@xxxxxxxx > Message-ID: <AUDITORY%201401270004144670.F12E@xxxxxxxx> > Precedence: list > List-Help: <http://lists.mcgill.ca/scripts/wa.exe?LIST=AUDITORY>, <mailto: > LISTSERV@xxxxxxxx?body=INFO AUDITORY> > List-Unsubscribe: <mailto:AUDITORY-unsubscribe-request@xxxxxxxx> > List-Subscribe: <mailto:AUDITORY-subscribe-request@xxxxxxxx> > List-Owner: <mailto:AUDITORY-request@xxxxxxxx> > List-Archive: <http://lists.mcgill.ca/scripts/wa.exe?LIST=AUDITORY> > > There are 2 messages totalling 1193 lines in this issue. > > Topics of the day: > > 1. Reference for typical SNRs is public spaces (2) > > Date: Sat, 25 Jan 2014 08:32:37 -0800 > From: Pierre Divenyi <pdivenyi@xxxxxxxx> > Subject: Re: Reference for typical SNRs is public spaces > MIME-Version: 1.0 > Content-Type: multipart/alternative; boundary="B_3473483561_12333698" > > > This message is in MIME format. Since your mail reader does not > understand > this format, some or all of this message may not be legible. > > --B_3473483561_12333698 > Content-type: text/plain; > charset="ISO-8859-1" > Content-transfer-encoding: quoted-printable > > I agree with Bill except for one (crucial) assumption: in a cocktail-party > situation the noise is not stationary (although, as Jont Allen once > suggested, adding talkers to the babble will make it approach > stationarity)= > . > So, the 0.5 dB SNR is workable in a broad statistical sense and would have > to be adjusted almost on a case-by-case basis. > > -Pierre > > From: "Richard F. Lyon" <dicklyon@xxxxxxxx> > Reply-To: "Richard F. Lyon" <dicklyon@xxxxxxxx> > Date: Friday, January 24, 2014 at 10:16 PM > To: <AUDITORY@xxxxxxxx> > Subject: Re: [AUDITORY] Reference for typical SNRs is public spaces > > I think Bill's point, which I agree with, is that the SNR is not determined > by the noise in the space as much as it is by people trying to communicate. > The SNR that he estimates is not "in the space", but rather "at the ears of > the listener" when the talker is trying to communicate to that particular > listener, above the noise. Other people trying to communicate to different > listeners make noise for this one. The SNR is therefore roughly constant, > somewhere near 0 dB, almost anywhere that's not too quiet. > > For me, it's a little higher, after I make people speak up. > > Dick > > > > On Fri, Jan 24, 2014 at 1:51 PM, Bill Woods <Bill_Woods@xxxxxxxx> > wrote: > > Dear List, > > =20 > > Since we=B9re assuming Andy is asking about speech-to-noise ratio when > refe= > rring > > to SNR in public spaces, I wondered if the question couldn=B9t be > answered = > by > > back-of-the-envelope calculations using some assumptions and articulation > > index theory (AI). > > =20 > > It turned out there was quite a long list of assumptions behind the > > calculation (see below), but the resultant simple calculation yielded a > S= > NR of > > 0.5 dB, which is quite in line with the values other posters have > indicat= > ed > > may actually be present. Two other facts became apparent as I did this, > > however, and further motivate my posting. > > =20 > > First, the long list of assumptions makes apparent the long list of > influ= > ences > > on the SNR in such a situation. These influences need to be > characterized= > in > > any attempt to generalize from SNR measurements in a given scenario, a > fa= > ct > > other posters have alluded to. > > =20 > > Second, it is important to note that much work has been done recently to > = > model > > or empirically characterize these influences on speech intelligibility, > a= > nd > > that, while the models may require more computational power than found > on= > the > > back of an envelope, they are no problem to execute on current laptop > > computers. The implication is that it should not be very difficult to > > determine a reasonably-accurate distribution of SNRs over wide variation > = > in > > the assumed listening scenario without any measurements. > > Cheers, > > Bill > > =20 > > Bill Woods, PhD > > Principal Research Scientist | Starkey Hearing Research Center > > 2150 Shattuck Ave. | Suite 408 | Berkeley, CA 94704-1345 > > T: 510-845-4876 x 14 <tel:510-845-4876%C2%A0x%C2%A014> > > starkey.com <http://www.starkey.com/> | starkeyresearch.com > > <http://www.starkeyresearch.com> | map > > < > http://maps.google.com/maps?q=3D2150+Shattuck+Avenue,+Berkeley,+CA+94704&h= > l=3Den& > > > ll=3D37.869941,-122.268219&spn=3D0.028152,0.038066&sll=3D37.86923,-122.273197&s= > spn=3D0 > > .056305,0.076132&z=3D15> | email <mailto:william_woods@xxxxxxxx> > > =20 > > The assumptions are: > > 1. Talkers have no hearing loss and no cognitive loss. > > 2. The talkers are facing each other and speaking in their =B3mother > tongue= > =B2. > > 3. We know the percent-correct (%C) targeted by the talkers and it is > les= > s > > than 100% (it=B9s a challenging environment). > > 4. We know the nature of the speech in such a conversation, from a > low-co= > ntext > > vs. high-context perspective. > > 5. The talkers are within their critical distance (i.e., ignore > reverbera= > tion > > of talkers=B9 speech). > > 6. We know the long-term spectral shape of speech at the eardrums. > > 7. We know the long-term spectral shape of noise at the eardrums. > > 8. The noise is stationary. > > 9. The noise is diffuse. > > 10. The diffuse noise in combination with diotic direct-wave target > speec= > h > > generates the equivalent of an =B3internal=B2 wideband binaural SNR > improveme= > nt of > > ~1.0 dB over monaural listening. > > 11. Overall level is not too high (i.e., no =B3roll-over=B2 effect for > > intelligibility has occurred). > > 12. We=B9re not including lip reading. > > =20 > > These assumptions allow one to, first, determine the AI needed to > achieve= > the > > assumed target %C given the assumed type of speech, and, second, > determin= > e the > > SNR required with the assumed spectral shapes to obtain that AI. > > =20 > > For instance, assuming talkers want 95%C with =B3unfamiliar sentences=B2 > then > > using the polynomial fits from Sherbecoe and Studebaker (JASA 1990) of > th= > e > > ANSI S3.5-1969 transfer functions between AI and %C, our talkers would > ne= > ed an > > AI of 0.45. If we assume the noise and speech have the same long-term > spe= > ctral > > shape then the SNR can be determined from (SNR+12)/30=3D0.45 (staying > with = > the > > 1969 AI method), yielding SNR =3D 1.5 dB. Subtracting the binaural SNR > > improvement yields 0.5 dB. > > =20 > > From: AUDITORY - Research in Auditory Perception > > [mailto:AUDITORY@xxxxxxxx On Behalf Of Andy Sabin > > Sent: Wednesday, January 22, 2014 9:53 AM > >=20 > > To: AUDITORY@xxxxxxxx > > Subject: Reference for typical SNRs is public spaces > > =20 > > Hi List,=20 > > =20 > > Can anyone point me to a reference showing SNRs that are typically > observ= > ed in > > public spaces (e.g., restaurants, bars ...etc)? I can find this info for > > overall SPL, but am having a hard time finding it for SNR. > > =20 > > Thanks > > Andy Sabin > > =20 > >=20 > >=20 > > > --B_3473483561_12333698 > Content-type: text/html; > charset="ISO-8859-1" > Content-transfer-encoding: quoted-printable > > <html><head></head><body style=3D"word-wrap: break-word; > -webkit-nbsp-mode: s= > pace; -webkit-line-break: after-white-space; color: rgb(0, 0, 0); > font-size:= > 14px; font-family: Calibri, sans-serif;"><div>I agree with Bill except > for = > one (crucial) assumption: in a cocktail-party situation the noise is not > sta= > tionary (although, as Jont Allen once suggested, adding talkers to the > babbl= > e will make it approach stationarity). So, the 0.5 dB SNR is workable in a > b= > road statistical sense and would have to be adjusted almost on a > case-by-cas= > e basis.</div><div> </div><div>-Pierre</div><div> </div> id=3D"OLK_= > SRC_BODY_SECTION"><div style=3D"font-family:Calibri; font-size:11pt; > text-alig= > n:left; color:black; BORDER-BOTTOM: medium none; BORDER-LEFT: medium none; > P= > ADDING-BOTTOM: 0in; PADDING-LEFT: 0in; PADDING-RIGHT: 0in; BORDER-TOP: > #b5c4= > df 1pt solid; BORDER-RIGHT: medium none; PADDING-TOP: 3pt"> style=3D"font= > -weight:bold">From: "Richard F. Lyon" <<a href=3D"mailto: > dicklyon@xxxxxxxx= > CM.ORG">dicklyon@xxxxxxxx</a>> style=3D"font-weight:bold">Reply-To:= > "Richard F. Lyon" <<a href=3D"mailto:dicklyon@xxxxxxxx > ">dicklyon@xxxxxxxx= > M.ORG</a>> Date: Friday, > Januar= > y 24, 2014 at 10:16 PM To: > <<a = > href=3D"mailto:AUDITORY@xxxxxxxx">AUDITORY@xxxxxxxx > </a>> <= > span style=3D"font-weight:bold">Subject: Re: [AUDITORY] Reference > for = > typical SNRs is public spaces </div><div> </div><div > dir=3D"ltr"><div>I = > think Bill's point, which I agree with, is that the SNR is not determined > by= > the noise in the space as much as it is by people trying to > communicate.&nb= > sp; The SNR that he estimates is not "in the space", but rather "at the > ears= > of the listener" when the talker is trying to communicate to that > particula= > r listener, above the noise.  Other people trying to communicate to > dif= > ferent listeners make noise for this one.  The SNR is therefore > roughly= > constant, somewhere near 0 dB, almost anywhere that's not too > quiet. <br= > ></div>For me, it's a little higher, after I make people speak > up. Di= > ck </div><div class=3D"gmail_extra"> <div > class=3D"gmail_quote">On= > Fri, Jan 24, 2014 at 1:51 PM, Bill Woods <<a > href=3D"mailt= > o:Bill_Woods@xxxxxxxx" target=3D"_blank">Bill_Woods@xxxxxxxx > </a>></sp= > an> wrote: <blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 > .8ex;borde= > r-left:1px #ccc solid;padding-left:1ex"><div> size=3D"3"><= > span style=3D"font-size:12pt"><div>Dear List,</div><div> face=3D"Times New = > Roman"> </div><div>Since we’re assuming Andy is asking > abo= > ut speech-to-noise ratio when referring to SNR in public spaces, I > wondered = > if the question couldn’t be answered by back-of-the-envelope > calculati= > ons using some assumptions and articulation index theory (AI).  > </div><= > div> </div><div>It turned out there was quite a long list of > assumption= > s behind the calculation (see below), but the resultant simple calculation > y= > ielded a SNR of 0.5 dB, which is quite in line with the values other > posters= > have indicated may actually be present. Two > other facts became apparent as I did this, however, and further motivate > my= > posting. </div><div> </div><div>First, the long list of assumptions > ma= > kes apparent the long list of influences on the SNR in such a situation. > The= > se influences need to be characterized in any attempt to generalize from > SNR= > measurements in a given scenario, a fact other posters have alluded > to.</div><div> </div><div style=3D"margin-bottom:12pt">Second, it is > imp= > ortant to note that much work has been done recently to model or > empirically= > characterize these influences on speech intelligibility, and that, while > th= > e models may require more computational power than found > on the back of an envelope, they are no problem to execute on current > lapto= > p computers. The implication is that it should not be very difficult to > dete= > rmine a reasonably-accurate distribution of SNRs over wide variation in > the = > assumed listening scenario without > any measurements. </div><div>Cheers,</div><div>Bill</div><div> face=3D"T= > imes New Roman"> </div><table > style=3D"width:282.75pt;margin-left:= > 5.4pt" width=3D"471"><colgroup><col style=3D"width:282.75pt" > width=3D"471"></colgr= > oup><tbody><tr><td>Bill Woods, > PhD <fo= > nt color=3D"#595959">Principal Research > Scientist= >  | Starkey Hearing Research Center > > 2150 Shattuck Ave. | Suite > 408 | Berkeley, CA&nbsp= > ;94704-1345 > > T: <a href=3D"tel:510-845-4876%C2%A0x%C2%A014" value=3D"+15108454876" > targ= > et=3D"_blank">510-845-4876 x 14</a> > </td></tr><= > tr><td><a href=3D"http://www.starkey.com/" > target= > =3D"_blank"> style=3D"font-size:10pt">st= > arkey.com</a> face=3D"Arial"> e=3D"font-size:10pt">  face=3D"Arial"><s= > pan style=3D"font-size:10pt">| face=3D"Arial"= > > <a href=3D" > http://www.starke= > yresearch.com" target=3D"_blank"> face=3D"Arial"> font-size:10pt">starkeyresearch.com</a> color=3D"#00= > 00F6" face=3D"Arial"> style=3D"font-size:10pt">  lor=3D"#595959" face=3D"Arial"> style=3D"font-size:10pt">|<font= > color=3D"#0000F6" face=3D"Arial"> style=3D"font-size:10pt"> </fo= > nt><a href=3D" > http://maps.google.com/maps?q=3D2150+Shattuck+Avenue,+Berkeley,+CA= > > +94704&hl=3Den&ll=3D37.869941,-122.268219&spn=3D0.028152,0.038066&= > sll=3D37.86923,-122.273197&sspn=3D0.056305,0.076132&z=3D15" > target=3D"_blank= > "> style=3D"font-size:10pt">map</s= > pan></a> style=3D"font-size:10p= > t">  style=3D"font-= > size:10pt">| style=3D"fo= > nt-size:10pt"> <a href=3D"mailto: > william_woods@xxxxxxxx"= > target=3D"_blank"> style=3D"font-size:10pt= > ">email</a></td></tr><tr><td> face=3D"Times N= > ew Roman"></td></tr><tr><td> Roman"></td>= > </tr></tbody></table><div> Roman"> </div><d= > iv>The assumptions are:</div><ol > style=3D"margin:0;padding-left:36pt"><li>Talk= > ers have no hearing loss and no cognitive loss.</li><li>The talkers are > faci= > ng each other and speaking in their “mother > tongue”.</li><li>We = > know the percent-correct (%C) targeted by the talkers and it is less than > 10= > 0% (it’s a challenging environment).</li><li>We know the nature of > the= > speech in such a conversation, from a low-context vs. high-context > perspect= > ive.</li><li>The talkers are within their critical distance (i.e., ignore > re= > verberation of talkers’ speech).</li><li> > We know the long-term spectral shape of speech at the eardrums.</li><li>We > = > know the long-term spectral shape of noise at the eardrums.</li><li>The > nois= > e is stationary.</li><li>The noise is diffuse.</li><li>The diffuse noise > in = > combination with diotic direct-wave target speech generates the equivalent > o= > f an “internal” wideband binaural SNR improvement of ~1.0 dB > ove= > r monaural listening.</li><li>Overall level is not too high (i.e., no > &#8220= > ;roll-over” effect for intelligibility has > occurred).</li><li>We&#8217= > ;re not including lip reading.</li></ol><div> Roman">&n= > bsp;</div><div>These assumptions allow one to, first, determine the > A= > I needed to achieve the assumed target %C given the assumed type of > speech, = > and, second, determine the SNR required with the assumed spectral shapes > to = > obtain that AI.</div><div> </div><div>For instance, assuming talkers > wa= > nt 95%C with “unfamiliar sentences” then using the polynomial > fi= > ts from Sherbecoe and Studebaker (JASA 1990) of the ANSI S3.5-1969 > transfer = > functions between AI and %C, our talkers would need an AI of 0.45. If we > ass= > ume > the noise and speech have the same long-term spectral shape then the SNR > ca= > n be determined from (SNR+12)/30=3D0.45 (staying with the 1969 AI method), > yie= > lding SNR =3D 1.5 dB.  Subtracting the binaural SNR improvement > yields 0.= > 5 dB.  </div><div> Roman"> </div><div>= > <div > class=3D"im">From:</b= > > AUDITORY - Research in Auditory Perception [<a href=3D"mailto: > AUDITORY@xxxxxxxx= > .MCGILL.CA" target=3D"_blank">mailto:AUDITORY@xxxxxxxx</a>] On > Behal= > f Of Andy Sabin </div>Sent: Wednesday, January 22, 2014 9:53 > A= > M<div class=3D"im"> To: <a href=3D"mailto: > AUDITORY@xxxxxxxx" ta= > rget=3D"_blank">AUDITORY@xxxxxxxx</a> Subject: Reference > for = > typical SNRs is public spaces</div></div><div > class=3D"im"><div>= >  </div><div> New R= > oman">Hi List, </div><div> Roman"> </f= > ont></div><div>Can anyone point me to a > referen= > ce showing SNRs that are typically observed in public spaces (e.g., > restaura= > nts, bars ...etc)? I can find this info for overall SPL, but am having a > har= > d time finding it for SNR. </div><div> Roma= > n"> </div><div> Roman">Thanks</div><= > div>Andy Sabin</div><div> face=3D"Tim= > es New > Roman"> </div></div></div></blockquote></di= > v> </div></body></html> > > --B_3473483561_12333698-- > > Date: Sun, 26 Jan 2014 09:54:31 +0000 > From: "Smits, Cas" <C.Smits@xxxxxxxx> > Subject: Re: Reference for typical SNRs is public spaces > MIME-Version: 1.0 > Content-Type: multipart/alternative; > boundary="_000_5801ADC15F4A8D4CAC8F4B3195FA6327EF3B322BSPMXMBX32vumcnl_" > > --_000_5801ADC15F4A8D4CAC8F4B3195FA6327EF3B322BSPMXMBX32vumcnl_ > Content-Type: text/plain; charset="us-ascii" > Content-Transfer-Encoding: quoted-printable > > We used a method to determine the relation between AI/SII and SNR for > fluct= > uating noise in a way quite similar to the way Bill proposed (Smits and > Fes= > ten, 2013). We found a different relationship between SII and SNR for > two-b= > and speech modulated noise compared to steady-state noise (a 2-dB increase > = > in SNR for fluctuating noise gives almost the same increase in SII as a > 1-d= > B increase in SNR for steady-state noise). This method works well in lab > se= > ttings but I think the uncertainties are too large to make reliable > predict= > ions for real-life situations. The effects of speaking style, speech > materi= > al, talker etc on percentage correct (the speech recognition function) are > = > large and the (backward) calculations give large uncertainties in SNR > estim= > ates. In situations with fluctuating noise the use of the steady-state > nois= > e relationship between SII and SNR could give large errors. With > interferin= > g talkers (informational masking) things get even more complicated. > > But, coming back to the question, normal-hearing listeners can follow most > = > conversations in public spaces with some difficulty (often in situations > wi= > th multitalker babble which is very similar to stationairy noise). Thus, > SN= > Rs around 0-5 dB SNR sounds plausible... > > Cas Smits > > > Smits C, Festen JM. The interpretation of speech reception threshold data > i= > n normal-hearing and hearing-impaired listeners: II. Fluctuating noise. > J Acoust Soc Am. 2013 May;133(5):3004-15 > > > dr.ir. J.C.M. (Cas) Smits, klinisch fysicus-audioloog | Universitair > Audiol= > ogisch Centrum | VU medisch centrum > Afdeling KNO-Audiologie | EMGO Institute for Health and Care Research > Secretariaat Audiologie, Postbus 7057, 1007 MB Amsterdam > T (020) 444 09 69 | F (020) 444 09 83 | E c.smits@xxxxxxxx<mailto: > c.smits@xxxxxxxx= > mc.nl> | W www.AC-VUmc.nl<http://www.AC-VUmc.nl> www.CI-VUmc.nl< > http://www= > .CI-VUmc.nl> > > > Disclaimer > Deze e-mail is vertrouwelijk en alleen bestemd voor de geadresseerde. Als > d= > eze e-mail niet voor u bestemd is, wordt u verzocht dit aan de afzender te > = > melden, het bericht niet verder te verspreiden en direct te vernietigen. > VU= > medisch centrum staat niet in voor de juiste en volledige overbrenging > van= > de inhoud van deze e-mail. > > > > Van: AUDITORY - Research in Auditory Perception [mailto: > AUDITORY@xxxxxxxx= > LL.CA] Namens Pierre Divenyi > Verzonden: Saturday, 25 January, 2014 17:33 > Aan: AUDITORY@xxxxxxxx > Onderwerp: Re: [AUDITORY] Reference for typical SNRs is public spaces > > I agree with Bill except for one (crucial) assumption: in a cocktail-party > = > situation the noise is not stationary (although, as Jont Allen once > suggest= > ed, adding talkers to the babble will make it approach stationarity). So, > t= > he 0.5 dB SNR is workable in a broad statistical sense and would have to > be= > adjusted almost on a case-by-case basis. > > -Pierre > > From: "Richard F. Lyon" <dicklyon@xxxxxxxx<mailto:dicklyon@xxxxxxxx>> > Reply-To: "Richard F. Lyon" <dicklyon@xxxxxxxx<mailto:dicklyon@xxxxxxxx>> > Date: Friday, January 24, 2014 at 10:16 PM > To: <AUDITORY@xxxxxxxx<mailto:AUDITORY@xxxxxxxx>> > Subject: Re: [AUDITORY] Reference for typical SNRs is public spaces > > I think Bill's point, which I agree with, is that the SNR is not > determined= > by the noise in the space as much as it is by people trying to > communicate= > . The SNR that he estimates is not "in the space", but rather "at the > ears= > of the listener" when the talker is trying to communicate to that > particul= > ar listener, above the noise. Other people trying to communicate to > differ= > ent listeners make noise for this one. The SNR is therefore roughly > consta= > nt, somewhere near 0 dB, almost anywhere that's not too quiet. > For me, it's a little higher, after I make people speak up. > > Dick > > On Fri, Jan 24, 2014 at 1:51 PM, Bill Woods <Bill_Woods@xxxxxxxx > <mailto:= > Bill_Woods@xxxxxxxx>> wrote: > Dear List, > > Since we're assuming Andy is asking about speech-to-noise ratio when > referr= > ing to SNR in public spaces, I wondered if the question couldn't be > answere= > d by back-of-the-envelope calculations using some assumptions and > articulat= > ion index theory (AI). > > It turned out there was quite a long list of assumptions behind the > calcula= > tion (see below), but the resultant simple calculation yielded a SNR of > 0.5= > dB, which is quite in line with the values other posters have indicated > ma= > y actually be present. Two other facts became apparent as I did this, > howev= > er, and further motivate my posting. > > First, the long list of assumptions makes apparent the long list of > influen= > ces on the SNR in such a situation. These influences need to be > characteriz= > ed in any attempt to generalize from SNR measurements in a given scenario, > = > a fact other posters have alluded to. > > Second, it is important to note that much work has been done recently to > mo= > del or empirically characterize these influences on speech > intelligibility,= > and that, while the models may require more computational power than > found= > on the back of an envelope, they are no problem to execute on current > lapt= > op computers. The implication is that it should not be very difficult to > de= > termine a reasonably-accurate distribution of SNRs over wide variation in > t= > he assumed listening scenario without any measurements. > Cheers, > Bill > > Bill Woods, PhD > Principal Research Scientist | Starkey Hearing Research Center > 2150 Shattuck Ave. | Suite 408 | Berkeley, CA 94704-1345 > T: 510-845-4876 x 14<tel:510-845-4876%C2%A0x%C2%A014> > > starkey.com<http://www.starkey.com/> | starkeyresearch.com< > http://www.stark= > eyresearch.com> | map< > http://maps.google.com/maps?q=3D2150+Shattuck+Avenue,= > > +Berkeley,+CA+94704&hl=3Den&ll=3D37.869941,-122.268219&spn=3D0.028152,0.038= > 066&sll=3D37.86923,-122.273197&sspn=3D0.056305,0.076132&z=3D15> | > email<mai= > lto:william_woods@xxxxxxxx> > > > > > The assumptions are: > 1. Talkers have no hearing loss and no cognitive loss. > 2. The talkers are facing each other and speaking in their "mother > ton= > gue". > 3. We know the percent-correct (%C) targeted by the talkers and it is > = > less than 100% (it's a challenging environment). > 4. We know the nature of the speech in such a conversation, from a > low= > -context vs. high-context perspective. > 5. The talkers are within their critical distance (i.e., ignore > reverb= > eration of talkers' speech). > 6. We know the long-term spectral shape of speech at the eardrums. > 7. We know the long-term spectral shape of noise at the eardrums. > 8. The noise is stationary. > 9. The noise is diffuse. > 10. The diffuse noise in combination with diotic direct-wave target > speech= > generates the equivalent of an "internal" wideband binaural SNR > improvemen= > t of ~1.0 dB over monaural listening. > 11. Overall level is not too high (i.e., no "roll-over" effect for > intelli= > gibility has occurred). > 12. We're not including lip reading. > > These assumptions allow one to, first, determine the AI needed to achieve > t= > he assumed target %C given the assumed type of speech, and, second, > determi= > ne the SNR required with the assumed spectral shapes to obtain that AI. > > For instance, assuming talkers want 95%C with "unfamiliar sentences" then > u= > sing the polynomial fits from Sherbecoe and Studebaker (JASA 1990) of the > A= > NSI S3.5-1969 transfer functions between AI and %C, our talkers would need > = > an AI of 0.45. If we assume the noise and speech have the same long-term > sp= > ectral shape then the SNR can be determined from (SNR+12)/30=3D0.45 > (stayin= > g with the 1969 AI method), yielding SNR =3D 1.5 dB. Subtracting the > binau= > ral SNR improvement yields 0.5 dB. > > From: AUDITORY - Research in Auditory Perception [mailto: > AUDITORY@xxxxxxxx= > ILL.CA] On Behalf Of Andy Sabin > Sent: Wednesday, January 22, 2014 9:53 AM > > To: AUDITORY@xxxxxxxx<mailto:AUDITORY@xxxxxxxx> > Subject: Reference for typical SNRs is public spaces > > Hi List, > > Can anyone point me to a reference showing SNRs that are typically > observed= > in public spaces (e.g., restaurants, bars ...etc)? I can find this info > fo= > r overall SPL, but am having a hard time finding it for SNR. > > Thanks > Andy Sabin > > > > --_000_5801ADC15F4A8D4CAC8F4B3195FA6327EF3B322BSPMXMBX32vumcnl_ > Content-Type: text/html; charset="us-ascii" > Content-Transfer-Encoding: quoted-printable > > <html xmlns:v=3D"urn:schemas-microsoft-com:vml" > xmlns:o=3D"urn:schemas-micr= > osoft-com:office:office" xmlns:w=3D"urn:schemas-microsoft-com:office:word" > = > xmlns:x=3D"urn:schemas-microsoft-com:office:excel" > xmlns:p=3D"urn:schemas-m= > icrosoft-com:office:powerpoint" > xmlns:a=3D"urn:schemas-microsoft-com:office= > :access" xmlns:dt=3D"uuid:C2F41010-65B3-11d1-A29F-00AA00C14882" > xmlns:s=3D"= > uuid:BDC6E3F0-6DA3-11d1-A2A3-00AA00C14882" > xmlns:rs=3D"urn:schemas-microsof= > t-com:rowset" xmlns:z=3D"#RowsetSchema" > xmlns:b=3D"urn:schemas-microsoft-co= > m:office:publisher" > xmlns:ss=3D"urn:schemas-microsoft-com:office:spreadshee= > t" xmlns:c=3D"urn:schemas-microsoft-com:office:component:spreadsheet" > xmlns= > :odc=3D"urn:schemas-microsoft-com:office:odc" > xmlns:oa=3D"urn:schemas-micro= > soft-com:office:activation" xmlns:html=3D"http://www.w3.org/TR/REC-html40" > = > xmlns:q=3D"http://schemas.xmlsoap.org/soap/envelope/" xmlns:rtc=3D" > http://m= > icrosoft.com/officenet/conferencing" xmlns:D=3D"DAV:" > xmlns:Repl=3D"http://= > schemas.microsoft.com/repl/" xmlns:mt=3D" > http://schemas.microsoft.com/share= > point/soap/meetings/" xmlns:x2=3D" > http://schemas.microsoft.com/office/excel= > /2003/xml" xmlns:ppda=3D"http://www.passport.com/NameSpace.xsd" xmlns:ois= > =3D"http://schemas.microsoft.com/sharepoint/soap/ois/" > xmlns:dir=3D"http://= > schemas.microsoft.com/sharepoint/soap/directory/" xmlns:ds=3D" > http://www.w3= > .org/2000/09/xmldsig#" xmlns:dsp=3D" > http://schemas.microsoft.com/sharepoint= > /dsp" xmlns:udc=3D"http://schemas.microsoft.com/data/udc" > xmlns:xsd=3D"http= > ://www.w3.org/2001/XMLSchema" xmlns:sub=3D" > http://schemas.microsoft.com/sha= > repoint/soap/2002/1/alerts/" xmlns:ec=3D"http://www.w3.org/2001/04/xmlenc# > "= > xmlns:sp=3D"http://schemas.microsoft.com/sharepoint/" > xmlns:sps=3D"http://= > schemas.microsoft.com/sharepoint/soap/" xmlns:xsi=3D" > http://www.w3.org/2001= > /XMLSchema-instance" xmlns:udcs=3D" > http://schemas.microsoft.com/data/udc/so= > ap" xmlns:udcxf=3D"http://schemas.microsoft.com/data/udc/xmlfile" > xmlns:udc= > p2p=3D"http://schemas.microsoft.com/data/udc/parttopart" > xmlns:m=3D"http://= > schemas.microsoft.com/office/2004/12/omml" xmlns:st=3D"" > xmlns=3D"http:= > //www.w3.org/TR/REC-html40"> > <head> > <meta http-equiv=3D"Content-Type" content=3D"text/html; > charset=3Dus-ascii"= > > > <meta name=3D"Generator" content=3D"Microsoft Word 12 (filtered medium)"> > <style> >  > </style> > </head> > <body lang=3D"EN-US" link=3D"blue" vlink=3D"purple" style=3D"word-wrap: > bre= > ak-word; > -webkit-nbsp-mode: space;-webkit-line-break: after-white-space"> > <div class=3D"Section1"> > style=3D"font-size:11.0pt;font-family:"Ca= > libri","sans-serif"; > color:#1F497D">We used a method to determine the relation between AI/SII > an= > d SNR for fluctuating noise in a way quite similar to the way Bill > proposed= > (Smits and Festen, > 2013). We found a different relationship between SII and SNR for two-band > = > speech modulated noise compared to steady-state noise (a 2-dB increase in > S= > NR for fluctuating noise gives almost the same increase in SII as a 1-dB > in= > crease in SNR for steady-state noise). > This method works well in lab settings but I think the uncertainties are > t= > oo large to make reliable predictions for real-life situations. The > effects= > of speaking style, speech material, talker etc on percentage correct (the > = > speech recognition function) are > large and the (backward) calculations give large uncertainties in SNR > esti= > mates. In situations with fluctuating noise the use of the steady-state > noi= > se relationship between SII and SNR could give large errors. With > interferi= > ng talkers (informational masking) > things get even more complicated.<o:p></o:p> > style=3D"font-size:11.0pt;font-family:"Ca= > libri","sans-serif"; > color:#1F497D"><o:p> </o:p> > style=3D"font-size:11.0pt;font-family:"Ca= > libri","sans-serif"; > color:#1F497D">But, coming back to the question, normal-hearing listeners > c= > an follow most conversations in public spaces with some difficulty (often > i= > n situations with multitalker > babble which is very similar to stationairy noise). Thus, SNRs around 0-5 > = > dB SNR sounds plausible...<o:p></o:p> > style=3D"font-size:11.0pt;font-family:"Ca= > libri","sans-serif"; > color:#1F497D"><o:p> </o:p> > style=3D"font-size:11.0pt;font-family:"Ca= > libri","sans-serif"; > color:#1F497D">Cas Smits<o:p></o:p> > style=3D"font-size:11.0pt;font-family:"Ca= > libri","sans-serif"; > color:#1F497D"><o:p> </o:p> > style=3D"font-size:11.0pt;font-family:"Ca= > libri","sans-serif"; > color:#1F497D"><o:p> </o:p> > style=3D"font-size:11.0pt;font-family:"Ca= > libri","sans-serif"; > color:#1F497D">Smits C, Festen JM. The interpretation of speech reception > t= > hreshold data in normal-hearing and hearing-impaired listeners: II. > Fluctua= > ting noise.<o:p></o:p> > style=3D"font-size:11.0pt;font-family:"Ca= > libri","sans-serif"; > color:#1F497D">J Acoust Soc Am. 2013 > May;133(5):3004-15<o:p></o:p></= > p> > style=3D"font-size:11.0pt;font-family:"Ca= > libri","sans-serif"; > color:#1F497D"><o:p> </o:p> > style=3D"font-size:11.0pt;font-family:"Ca= > libri","sans-serif"; > color:#1F497D"><o:p> </o:p> > style=3D"line-height:12.0pt;mso-line-height-rule:exa= > ctly"> style=3D"font-size:8.0pt;font-family:"Verdana",&qu= > ot;sans-serif";color:#10137D">dr.ir. J.C.M. (Cas) Smits, klinisch > fysi= > cus-audioloog style=3D"font-size:8.0pt;font-family:ver;c= > olor:#10137D">  style=3D"font-size:8.0pt;font-famil= > y:"Verdana","sans-serif";color:#10137D">| > Universitair Audiologisch Centrum | VU medisch centrum sty= > le=3D"font-size: > 8.0pt;font-family:"Verdana","sans-serif";color:black"> > > style=3D"font-size:8.0pt;font-family:"Verdana",&quot= > ;sans-serif"; > color:black">Afdeling KNO-Audiologie | EMGO Institute for Health and Care > R= > esearch<o:p></o:p> > style=3D"line-height:12.0pt;mso-line-height-rule:exa= > ctly"> style=3D"font-size:8.0pt;font-family:"Verdana&= > quot;,"sans-serif";color:black">Secretariaat Audiologie, Postbus > = > 7057, 1007 MB Amsterdam > style=3D"font-size:8.0pt;font-family:"Verd= > ana","sans-serif"; > color:#E5053A">T style=3D"font-size:8.0pt;font-= > family: > "Verdana","sans-serif";color:black"> (020) 444 09 69 > |&= > nbsp; style=3D"font-size:8.0pt;font-family:&quot= > ;Verdana","sans-serif";color:#E5053A">F lang= > =3D"NL" > style=3D"font-size:8.0pt;font-family:"Verdana","sans= > -serif";color:black"> > (020) 444 09 83 |  style=3D"font-size:8.0p= > t; > > font-family:"Verdana","sans-serif";color:#E5053A">E</sp= > an> style=3D"font-size:8.0pt;font-family:"Verdana= > ","sans-serif";color:black">  style=3D"colo= > r:#10137D"><a href=3D"mailto:c.smits@xxxxxxxx"> style=3D"fo= > > nt-size:8.0pt;font-family:"Verdana","sans-serif"">c.smi= > ts@xxxxxxxx</a> style=3D"font-size:8.0pt;font= > -family:"Verdana","sans-serif";color:black"> | > style=3D"font-size:8.0pt;font-family:"Verd= > ana","sans-serif"; > color:#E5053A">W style=3D"font-size:8.0pt;font-= > family: > "Verdana","sans-serif";color:black"> > <a href=3D"http://www.AC-VUmc.nl > "><spa= > n lang=3D"NL" style=3D"font-size:8.0pt;font-family: > "Verdana","sans-serif"">www.AC-VUmc.nl > </a></span= > > style=3D"font-size:8.0pt;font-family:"Verdana"= > ,"sans-serif";color:black">  > =3D"font-size:8.0pt;font-family:"Verdana","sans-serif";= > color:#E5053A">  style=3D"font-size:8.0pt;font-family:&quo= > t;Verdana","sans-serif";color:purple"><a href=3D" > http://www.= > CI-VUmc.nl">www.CI-VUmc.nl</a> lang=3D= > "NL" style=3D"color:#10137D"><o:p></o:p> > style=3D"line-height:12.0pt;mso-line-height-rule:exa= > ctly"> style=3D"color:#10137D"> <o:p></o:p></p= > > > style=3D"color:#10137D"> <o:p= > ></o:p> > style=3D"font-size:9.0pt;color:= > gray">Disclaimer<o:p></o:p> > style=3D"font-size:9.0pt;color:gra= > y">Deze e-mail is vertrouwelijk en alleen bestemd voor de geadresseerde. > Al= > s deze e-mail niet voor u bestemd is, wordt u verzocht dit aan de afzender > = > te melden, het bericht niet verder te > verspreiden en direct te vernietigen. VU medisch centrum staat niet in > voo= > r de juiste en volledige overbrenging van de inhoud van deze > e-mail.= > <o:p></o:p> > style=3D"font-size:11.0pt;font-fam= > ily:"Calibri","sans-serif"; > color:#1F497D"><o:p> </o:p> > style=3D"font-size:11.0pt;font-fam= > ily:"Calibri","sans-serif"; > color:#1F497D"><o:p> </o:p> > style=3D"font-size:11.0pt;font-fam= > ily:"Calibri","sans-serif"; > color:#1F497D"><o:p> </o:p> > <div> > <div style=3D"border:none;border-top:solid #B5C4DF 1.0pt;padding:3.0pt 0cm > = > 0cm 0cm"> > style=3D"font-size:10.0pt;font-= > family:"Tahoma","sans-serif"">Van: lang= > =3D"NL" > style=3D"font-size:10.0pt;font-family:"Tahoma","sans= > -serif""> AUDITORY - Research in Auditory Perception [mailto: > AUDITORY@xxxxxxxx= > LISTS.MCGILL.CA] > Namens Pierre Divenyi > Verzonden: Saturday, 25 January, 2014 17:33 > Aan: AUDITORY@xxxxxxxx > Onderwerp: Re: [AUDITORY] Reference for typical SNRs is public > space= > s<o:p></o:p> > </div> > </div> > <o:p> </o:p> > <div> > style=3D"font-size:10.5pt;font-family:"Ca= > libri","sans-serif"; > color:black">I agree with Bill except for one (crucial) assumption: in a > co= > cktail-party situation the noise is not stationary (although, as Jont > Allen= > once suggested, adding > talkers to the babble will make it approach stationarity). So, the 0.5 dB > = > SNR is workable in a broad statistical sense and would have to be adjusted > = > almost on a case-by-case basis.<o:p></o:p> > </div> > <div> > style=3D"font-size:10.5pt;font-family:"Ca= > libri","sans-serif"; > color:black"><o:p> </o:p> > </div> > <div> > style=3D"font-size:10.5pt;font-family:"Ca= > libri","sans-serif"; > color:black">-Pierre<o:p></o:p> > </div> > <div> > style=3D"font-size:10.5pt;font-family:"Ca= > libri","sans-serif"; > color:black"><o:p> </o:p> > </div> > <div style=3D"border:none;border-top:solid #B5C4DF 1.0pt;padding:3.0pt 0cm > = > 0cm 0cm"> > style=3D"font-size:11.0pt;font-family:&quot= > ;Calibri","sans-serif"; > color:black">From: > style=3D"font-size:11.0pt;font-family:"Calibri",= > "sans-serif"; > color:black">"Richard F. Lyon" <<a href=3D"mailto: > dicklyon@xxxxxxxx= > .ORG">dicklyon@xxxxxxxx</a>> > Reply-To: "Richard F. Lyon" <<a href=3D"mailto: > dicklyon= > @xxxxxxxx">dicklyon@xxxxxxxx</a>> > Date: Friday, January 24, 2014 at 10:16 PM > To: <<a href=3D"mailto:AUDITORY@xxxxxxxx > ">AUDITORY@xxxxxxxx= > CGILL.CA</a>> > Subject: Re: [AUDITORY] Reference for typical SNRs is public > spaces<= > o:p></o:p> > </div> > <div> > style=3D"font-size:10.5pt;font-family:"Ca= > libri","sans-serif"; > color:black"><o:p> </o:p> > </div> > <div> > <div> > style=3D"font-s= > ize:10.5pt; > font-family:"Calibri","sans-serif";color:black">I > think= > Bill's point, which I agree with, is that the SNR is not determined by > the= > noise in the space as much as it is by people > trying to communicate.  The SNR that he estimates is not "in > the= > space", but rather "at the ears of the listener" when the > t= > alker is trying to communicate to that particular listener, above the > noise= > .  Other people trying to communicate to different listeners > make noise for this one.  The SNR is therefore roughly constant, > some= > where near 0 dB, almost anywhere that's not too > quiet.<o:p></o:p></p= > > > </div> > style=3D"font-s= > ize:10.5pt; > font-family:"Calibri","sans-serif";color:black">For > me,= > it's a little higher, after I make people speak up. > > Dick<o:p></o:p> > </div> > <div> > style=3D"font-s= > ize:10.5pt; > > font-family:"Calibri","sans-serif";color:black"><o:p>&n= > bsp;</o:p> > <div> > style=3D"font-size:10.5pt;font-family:"Ca= > libri","sans-serif"; > color:black">On Fri, Jan 24, 2014 at 1:51 PM, Bill Woods <<a > href=3D"mai= > lto:Bill_Woods@xxxxxxxx" target=3D"_blank">Bill_Woods@xxxxxxxx > </a>&gt= > ; wrote:<o:p></o:p> > <div> > <div> > style=3D"font-family:"Calibri",&quot= > ;sans-serif";color:black">Dear List,<o:p></o:p> > </div> > <div> >   style= > > =3D"font-family:"Calibri","sans-serif";color:black"><o:= > p></o:p> > </div> > <div> > style=3D"font-family:"Calibri",&quot= > ;sans-serif";color:black">Since we’re assuming Andy is asking > ab= > out speech-to-noise ratio when referring to SNR in public spaces, I > wondere= > d if the question couldn’t be answered by back-of-the-envelope > calculations using some assumptions and articulation index theory > (AI).&nb= > sp; <o:p></o:p> > </div> > <div> > style=3D"font-family:"Calibri",&quot= > ;sans-serif";color:black"> <o:p></o:p> > </div> > <div> > style=3D"font-family:"Calibri",&quot= > ;sans-serif";color:black">It turned out there was quite a long list > of= > assumptions behind the calculation (see below), but the resultant simple > c= > alculation yielded a SNR of 0.5 dB, which is quite in line > with the values other posters have indicated may actually be present. Two > = > other facts became apparent as I did this, however, and further motivate > my= > posting. > <o:p></o:p> > </div> > <div> > style=3D"font-family:"Calibri",&quot= > ;sans-serif";color:black"> <o:p></o:p> > </div> > <div> > style=3D"font-family:"Calibri",&quot= > ;sans-serif";color:black">First, the long list of assumptions makes > ap= > parent the long list of influences on the SNR in such a situation. These > in= > fluences need to be characterized in any attempt to generalize > from SNR measurements in a given scenario, a fact other posters have > allud= > ed to.<o:p></o:p> > </div> > <div> > style=3D"font-family:"Calibri",&quot= > ;sans-serif";color:black"> <o:p></o:p> > </div> > <div style=3D"margin-bottom:12.0pt"> > style=3D"font-family:"Calibri",&quot= > ;sans-serif";color:black">Second, it is important to note that much > wo= > rk has been done recently to model or empirically characterize these > influe= > nces on speech intelligibility, and that, while the models > may require more computational power than found on the back of an > envelope= > , they are no problem to execute on current laptop computers. The > implicati= > on is that it should not be very difficult to determine a > reasonably-accura= > te distribution of SNRs over wide > variation in the assumed listening scenario without any measurements. > <o:p= > ></o:p> > </div> > <div> > style=3D"font-family:"Calibri",&quot= > ;sans-serif";color:black">Cheers,<o:p></o:p> > </div> > <div> > style=3D"font-family:"Calibri",&quot= > ;sans-serif";color:black">Bill<o:p></o:p> > </div> > <div> >   style= > > =3D"font-family:"Calibri","sans-serif";color:black"><o:= > p></o:p> > </div> > <table class=3D"MsoNormalTable" border=3D"0" cellpadding=3D"0" > width=3D"377= > " style=3D"width:282.75pt; > margin-left:5.4pt"> > <tbody> > <tr> > <td style=3D"padding:.75pt .75pt .75pt .75pt"> > style=3D"font-family:"Arial","s= > ans-serif";color:#17365D">Bill Woods, PhD > style=3D"font-size:10.0pt;font-family:"Arial","= > sans-serif"; > color:#595959">Principal Research Scientist | Starkey Hearing > R= > esearch Center > 2150 Shattuck Ave. | Suite > 408 | Berkeley, CA&nbsp= > ;94704-1345 > T: <a href=3D"tel:510-845-4876%C2%A0x%C2%A014" > target=3D"_blank">510-8= > 45-4876 x 14</a> > <o:p></o:p> > </td> > </tr> > <tr> > <td style=3D"padding:.75pt .75pt .75pt .75pt"> > <a href=3D"http://www.starkey.com/" > target=3D"_blank= > "> style=3D"font-size:10.0pt;font-family:"Arial","sans-= > serif"">starkey.com</a> style=3D"font-size:10.0pt;font-fam= > > ily:"Arial","sans-serif";color:#0000F6"> <s= > pan > style=3D"font-size:10.0pt;font-family:"Arial","sans-seri= > f";color:#595959">| style=3D"font-size:10.0pt;font-family:= > "Arial","sans-serif";color:#0000F6"> <a > hre= > f=3D"http://www.starkeyresearch.com" target=3D"_blank"> style=3D"font-= > size: > > 10.0pt;font-family:"Arial","sans-serif"">starkeyresea= > rch.com</a> style=3D"font-size:10.0pt;font-family:"Arial&q= > uot;,"sans-serif";color:#0000F6">  style=3D"font= > > -size:10.0pt;font-family:"Arial","sans-serif";color:#59= > 5959">| style=3D"font-size:10.0pt;font-family:"Arial"= > ,"sans-serif";color:#0000F6"> <a href=3D" > http://maps.= > > google.com/maps?q=3D2150+Shattuck+Avenue,+Berkeley,+CA+= > > 94704&hl=3Den&ll=3D37.869941,-122.268219&spn=3D0.028152,0.03806= > 6&sll=3D37.86923,-122.273197&sspn=3D0.056305,0.076132&z=3D15" > t= > arget=3D"_blank"> style=3D"font-size:10.0pt;font-family:"Arial&qu= > ot;,"sans-serif"">map</a> style=3D"font-size:10.0pt;f= > > ont-family:"Arial","sans-serif";color:#0000F6"> </= > span> style=3D"font-size:10.0pt;font-family:"Arial","sa= > ns-serif";color:#595959">| style=3D"font-size:10.0pt;font-= > > family:"Arial","sans-serif";color:#0000F6"> </span= > ><a href=3D"mailto:william_woods@xxxxxxxx" target=3D"_blank"> style= > > =3D"font-size:10.0pt;font-family:"Arial","sans-serif"">= > email</a><o:p></o:p> > </td> > </tr> > <tr> > <td style=3D"padding:.75pt .75pt .75pt .75pt"></td> > </tr> > <tr> > <td style=3D"padding:.75pt .75pt .75pt .75pt"></td> > </tr> > </tbody> > </table> > <div> >   style= > > =3D"font-family:"Calibri","sans-serif";color:black"><o:= > p></o:p> > </div> > <div> > style=3D"font-family:"Calibri",&quot= > ;sans-serif";color:black">The assumptions are:<o:p></o:p> > </div> > style=3D"mso-margin-top-alt:auto;mso-margin-bottom-a= > lt:auto; > margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> > <![if !supportLists]> style=3D"font-family:"Calibri","s= > ans-serif";color:black"> Ignore">1. Roman""> &nbs= > p;    > <![endif]> style=3D"font-family:"Calibri&quo= > t;,"sans-serif"; > color:black">Talkers have no hearing loss and no cognitive > loss.<o:p></o:p>= > > style=3D"mso-margin-top-alt:auto;mso-margin-bottom-a= > lt:auto; > margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> > <![if !supportLists]> style=3D"font-family:"Calibri","s= > ans-serif";color:black"> Ignore">2. Roman""> &nbs= > p;    > <![endif]> style=3D"font-family:"Calibri&quo= > t;,"sans-serif"; > color:black">The talkers are facing each other and speaking in their > &#8220= > ;mother tongue”.<o:p></o:p> > style=3D"mso-margin-top-alt:auto;mso-margin-bottom-a= > lt:auto; > margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> > <![if !supportLists]> style=3D"font-family:"Calibri","s= > ans-serif";color:black"> Ignore">3. Roman""> &nbs= > p;    > <![endif]> style=3D"font-family:"Calibri&quo= > t;,"sans-serif"; > color:black">We know the percent-correct (%C) targeted by the talkers and > i= > t is less than 100% (it’s a challenging > environment).<o:p></o:p></spa= > n> > style=3D"mso-margin-top-alt:auto;mso-margin-bottom-a= > lt:auto; > margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> > <![if !supportLists]> style=3D"font-family:"Calibri","s= > ans-serif";color:black"> Ignore">4. Roman""> &nbs= > p;    > <![endif]> style=3D"font-family:"Calibri&quo= > t;,"sans-serif"; > color:black">We know the nature of the speech in such a conversation, from > = > a low-context vs. high-context perspective.<o:p></o:p> > style=3D"mso-margin-top-alt:auto;mso-margin-bottom-a= > lt:auto; > margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> > <![if !supportLists]> style=3D"font-family:"Calibri","s= > ans-serif";color:black"> Ignore">5. Roman""> &nbs= > p;    > <![endif]> style=3D"font-family:"Calibri&quo= > t;,"sans-serif"; > color:black">The talkers are within their critical distance (i.e., ignore > r= > everberation of talkers’ speech).<o:p></o:p> > style=3D"mso-margin-top-alt:auto;mso-margin-bottom-a= > lt:auto; > margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> > <![if !supportLists]> style=3D"font-family:"Calibri","s= > ans-serif";color:black"> Ignore">6. Roman""> &nbs= > p;    > <![endif]> style=3D"font-family:"Calibri&quo= > t;,"sans-serif"; > color:black">We know the long-term spectral shape of speech at the > eardrums= > .<o:p></o:p> > style=3D"mso-margin-top-alt:auto;mso-margin-bottom-a= > lt:auto; > margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> > <![if !supportLists]> style=3D"font-family:"Calibri","s= > ans-serif";color:black"> Ignore">7. Roman""> &nbs= > p;    > <![endif]> style=3D"font-family:"Calibri&quo= > t;,"sans-serif"; > color:black">We know the long-term spectral shape of noise at the > eardrums.= > <o:p></o:p> > style=3D"mso-margin-top-alt:auto;mso-margin-bottom-a= > lt:auto; > margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> > <![if !supportLists]> style=3D"font-family:"Calibri","s= > ans-serif";color:black"> Ignore">8. Roman""> &nbs= > p;    > <![endif]> style=3D"font-family:"Calibri&quo= > t;,"sans-serif"; > color:black">The noise is stationary.<o:p></o:p> > style=3D"mso-margin-top-alt:auto;mso-margin-bottom-a= > lt:auto; > margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> > <![if !supportLists]> style=3D"font-family:"Calibri","s= > ans-serif";color:black"> Ignore">9. Roman""> &nbs= > p;    > <![endif]> style=3D"font-family:"Calibri&quo= > t;,"sans-serif"; > color:black">The noise is diffuse.<o:p></o:p> > style=3D"mso-margin-top-alt:auto;mso-margin-bottom-a= > lt:auto; > margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> > <![if !supportLists]> style=3D"font-family:"Calibri","s= > ans-serif";color:black"> Ignore">10.  > <![endif]> style=3D"font-family:"Calibri&quo= > t;,"sans-serif";color:black">The diffuse noise in combination > wit= > h diotic direct-wave target speech generates the equivalent of an > “in= > ternal” wideband binaural SNR improvement of ~1.0 dB over monaural > listening.<o:p></o:p> > style=3D"mso-margin-top-alt:auto;mso-margin-bottom-a= > lt:auto; > margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> > <![if !supportLists]> style=3D"font-family:"Calibri","s= > ans-serif";color:black"> Ignore">11.  > <![endif]> style=3D"font-family:"Calibri&quo= > t;,"sans-serif";color:black">Overall level is not too high > (i.e.,= > no “roll-over” effect for intelligibility has > occurred).<o:p><= > /o:p> > style=3D"mso-margin-top-alt:auto;mso-margin-bottom-a= > lt:auto; > margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> > <![if !supportLists]> style=3D"font-family:"Calibri","s= > ans-serif";color:black"> Ignore">12.  > <![endif]> style=3D"font-family:"Calibri&quo= > t;,"sans-serif";color:black">We’re not including lip > readin= > g.<o:p></o:p> > <div> >   style= > > =3D"font-family:"Calibri","sans-serif";color:black"><o:= > p></o:p> > </div> > <div> > style=3D"font-family:"Calibri",&quot= > ;sans-serif";color:black">These assumptions allow one to, first, > deter= > mine the AI needed to achieve the assumed target %C given the assumed type > = > of speech, and, second, determine the SNR required with > the assumed spectral shapes to obtain that AI.<o:p></o:p> > </div> > <div> > style=3D"font-family:"Calibri",&quot= > ;sans-serif";color:black"> <o:p></o:p> > </div> > <div> > style=3D"font-family:"Calibri",&quot= > ;sans-serif";color:black">For instance, assuming talkers want 95%C > wit= > h “unfamiliar sentences” then using the polynomial fits from > Sh= > erbecoe and Studebaker (JASA 1990) of the ANSI S3.5-1969 transfer functions > between AI and %C, our talkers would need an AI of 0.45. If we assume the > = > noise and speech have the same long-term spectral shape then the SNR can > be= > determined from (SNR+12)/30=3D0.45 (staying with the 1969 AI method), > = > yielding SNR =3D 1.5 dB.  Subtracting the > binaural SNR improvement yields 0.5 dB.  <o:p></o:p> > </div> > <div> >   style= > > =3D"font-family:"Calibri","sans-serif";color:black"><o:= > p></o:p> > </div> > <div> > <div> > style=3D"font-size:10.0pt;font-family:&quot= > ;Tahoma","sans-serif"; > color:black">From: style=3D"font-size:10.0pt;font-family:&q= > uot;Tahoma","sans-serif"; > color:black"> AUDITORY - Research in Auditory Perception [<a > href=3D"mailto= > :AUDITORY@xxxxxxxx" target=3D"_blank">mailto: > AUDITORY@xxxxxxxx= > A</a>] > On Behalf Of Andy Sabin<o:p></o:p> > </div> > style=3D"font-size:10.0pt;font-family:&quot= > ;Tahoma","sans-serif"; > color:black">Sent: style=3D"font-size:10.0pt;font-family:&q= > uot;Tahoma","sans-serif"; > color:black"> Wednesday, January 22, 2014 9:53 AM<o:p></o:p> > <div> > style=3D"font-size:10.0pt;font-family:"Ta= > homa","sans-serif"; > color:black"> > To: <a href=3D"mailto:AUDITORY@xxxxxxxx" > target=3D"_blank">AU= > DITORY@xxxxxxxx</a> > Subject: Reference for typical SNRs is public > spaces<o:p></o:p></spa= > n> > </div> > </div> > <div> > <div> >   style= > > =3D"font-family:"Calibri","sans-serif";color:black"><o:= > p></o:p> > </div> > <div> > Hi > List, <sp= > an > style=3D"font-family:"Calibri","sans-serif";color:bl= > ack"><o:p></o:p> > </div> > <div> >   style= > > =3D"font-family:"Calibri","sans-serif";color:black"><o:= > p></o:p> > </div> > <div> > Can anyone point me to > a= > reference showing SNRs that are typically observed in public spaces > (e.g.,= > restaurants, bars ...etc)? I can find this info for overall SPL, but am > ha= > ving a hard time finding it for SNR.  style=3D"font-family= > :"Calibri","sans-serif"; > color:black"><o:p></o:p> > </div> > <div> >   style= > > =3D"font-family:"Calibri","sans-serif";color:black"><o:= > p></o:p> > </div> > <div> > Thanks style= > > =3D"font-family:"Calibri","sans-serif";color:black"><o:= > p></o:p> > </div> > <div> > Andy Sabin s= > > tyle=3D"font-family:"Calibri","sans-serif";color:black"= > ><o:p></o:p> > </div> > <div> >   style= > > =3D"font-family:"Calibri","sans-serif";color:black"><o:= > p></o:p> > </div> > </div> > </div> > </div> > style=3D"font-size:10.5pt;font-family:"Ca= > libri","sans-serif"; > color:black"><o:p> </o:p> > </div> > </div> > </body> > </html> > > --_000_5801ADC15F4A8D4CAC8F4B3195FA6327EF3B322BSPMXMBX32vumcnl_-- > > --bcaec548a62757d44904f1009bf3 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable <div dir=3D"ltr">Many thanks to all for super thoughtful responses! For cla= rification, I was interested in typical SNRs for going out to a restaurant/= bar. A few folks have asked for a summary... Seems like there are two main = points: (1) there are many sources of variation in the talker/listener/envi= ronment that can have an enormous effect on SNR, and (2) there is a princip= led reason to think that, on average, the SNR often ends up near 0 dB=A0(in= situations that are limited by noise)=A0. =A0<div> </div><div>There might be a fun informal experiment you can try the nex= t time you are at a restaurant/bar: Measure the SPL near your mouth when ta= lking (there are many free apps that are decent), estimate the level of you= r voice at the listener's ear using the mouth-to-ear distance, and comp= are that level to the SPL at the listener's ear when you are not talkin= g. If you do that, I'd be interested to know what you find (off list pl= ease).=A0<div> <div> </div></div></div></div><div class=3D"gmail_extra"> <div class= =3D"gmail_quote">On Sun, Jan 26, 2014 at 11:04 PM, AUDITORY automatic diges= t system <<a href=3D"mailto:LISTSERV@xxxxxxxx" = target=3D"_blank">LISTSERV@xxxxxxxx</a>> wrote: <blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p= x #ccc solid;padding-left:1ex">Date: =A0 =A0 Mon, 27 Jan 2014 00:04:14 -050= 0 Reply-To: AUDITORY - Research in Auditory Perception <<a href=3D"mailto:= AUDITORY@xxxxxxxx">AUDITORY@xxxxxxxx</a>> Sender: =A0 AUDITORY - Research in Auditory Perception <<a href=3D"mailt= o:AUDITORY@xxxxxxxx">AUDITORY@xxxxxxxx</a>> From: =A0 =A0 AUDITORY automatic digest system <<a href=3D"mailto:LISTSE= RV@xxxxxxxx">LISTSERV@xxxxxxxx</a>> Subject: AUDITORY Digest - 25 Jan 2014 to 26 Jan 2014 (#2014-29) To: =A0 =A0 =A0 <a href=3D"mailto:AUDITORY@xxxxxxxx">AUDITORY@xxxxxxxx= MCGILL.CA</a> Message-ID: <<a href=3D"mailto:AUDITORY%25201401270004144670.F12E@xxxxxxxx= MCGILL.CA">AUDITORY%201401270004144670.F12E@xxxxxxxx</a>> Precedence: list List-Help: <<a href=3D"http://lists.mcgill.ca/scripts/wa.exe?LIST=3DAUDI= TORY" target=3D"_blank">http://lists.mcgill.ca/scripts/wa.exe?LIST=3DAUDITO= RY</a>>, <mailto:<a href=3D"mailto:LISTSERV@xxxxxxxx">LISTSERV= @xxxxxxxx</a>?body=3DINFO AUDITORY> List-Unsubscribe: <mailto:<a href=3D"mailto:AUDITORY-unsubscribe-request= @xxxxxxxx">AUDITORY-unsubscribe-request@xxxxxxxx</a>> List-Subscribe: <mailto:<a href=3D"mailto:AUDITORY-subscribe-request@xxxxxxxx= TS.MCGILL.CA">AUDITORY-subscribe-request@xxxxxxxx</a>> List-Owner: <mailto:<a href=3D"mailto:AUDITORY-request@xxxxxxxx">= AUDITORY-request@xxxxxxxx</a>> List-Archive: <<a href=3D"http://lists.mcgill.ca/scripts/wa.exe?LIST=3DA= UDITORY" target=3D"_blank">http://lists.mcgill.ca/scripts/wa.exe?LIST=3DAUD= ITORY</a>> There are 2 messages totalling 1193 lines in this issue. Topics of the day: =A0 1. Reference for typical SNRs is public spaces (2) Date: =A0 =A0Sat, 25 Jan 2014 08:32:37 -0800 From: =A0 =A0Pierre Divenyi <<a href=3D"mailto:pdivenyi@xxxxxxxx= DU">pdivenyi@xxxxxxxx</a>> Subject: Re: Reference for typical SNRs is public spaces MIME-Version: 1.0 Content-Type: multipart/alternative; boundary=3D"B_<a href=3D"tel:3473= 483561" value=3D"+13473483561">3473483561</a>_12333698" > This message is in MIME format. Since your mail reader does not unders= tand this format, some or all of this message may not be legible. --B_<a href=3D"tel:3473483561" value=3D"+13473483561">3473483561</a>_123336= 98 Content-type: text/plain; =A0 =A0 =A0 =A0 charset=3D"ISO-8859-1" Content-transfer-encoding: quoted-printable I agree with Bill except for one (crucial) assumption: in a cocktail-party<= br> situation the noise is not stationary (although, as Jont Allen once suggested, adding talkers to the babble will make it approach stationarity)= =3D . So, the 0.5 dB SNR is workable in a broad statistical sense and would have<= br> to be adjusted almost on a case-by-case basis. -Pierre From: =A0"Richard F. Lyon" <<a href=3D"mailto:dicklyon@xxxxxxxx= ">dicklyon@xxxxxxxx</a>> Reply-To: =A0"Richard F. Lyon" <<a href=3D"mailto:dicklyon@xxxxxxxx= .ORG">dicklyon@xxxxxxxx</a>> Date: =A0Friday, January 24, 2014 at 10:16 PM To: =A0<<a href=3D"mailto:AUDITORY@xxxxxxxx">AUDITORY@xxxxxxxx= L.CA</a>> Subject: =A0Re: [AUDITORY] Reference for typical SNRs is public spaces I think Bill's point, which I agree with, is that the SNR is not determ= ined by the noise in the space as much as it is by people trying to communicate.= The SNR that he estimates is not "in the space", but rather &quot= ;at the ears of the listener" when the talker is trying to communicate to that particu= lar listener, above the noise. =A0Other people trying to communicate to differe= nt listeners make noise for this one. =A0The SNR is therefore roughly constant= , somewhere near 0 dB, almost anywhere that's not too quiet. For me, it's a little higher, after I make people speak up. Dick On Fri, Jan 24, 2014 at 1:51 PM, Bill Woods <<a href=3D"mailto:Bill_Wood= s@xxxxxxxx">Bill_Woods@xxxxxxxx</a>> wrote: > Dear List, > =3D20 > Since we=3DB9re assuming Andy is asking about speech-to-noise ratio wh= en refe=3D rring > to SNR in public spaces, I wondered if the question couldn=3DB9t be an= swered =3D by > back-of-the-envelope calculations using some assumptions and articulat= ion > index theory (AI). > =3D20 > It turned out there was quite a long list of assumptions behind the<br= > > calculation (see below), but the resultant simple calculation yielded = a S=3D NR of > 0.5 dB, which is quite in line with the values other posters have indi= cat=3D ed > may actually be present. Two other facts became apparent as I did this= , > however, and further motivate my posting. > =3D20 > First, the long list of assumptions makes apparent the long list of in= flu=3D ences > on the SNR in such a situation. These influences need to be characteri= zed=3D =A0in > any attempt to generalize from SNR measurements in a given scenario, a= fa=3D ct > other posters have alluded to. > =3D20 > Second, it is important to note that much work has been done recently = to =3D model > or empirically characterize these influences on speech intelligibility= , a=3D nd > that, while the models may require more computational power than found= on=3D =A0the > back of an envelope, they are no problem to execute on current laptop<= br> > computers. The implication is that it should not be very difficult to<= br> > determine a reasonably-accurate distribution of SNRs over wide variati= on =3D in > the assumed listening scenario without any measurements. > Cheers, > Bill > =3D20 > Bill Woods, PhD > Principal Research Scientist | Starkey Hearing Research Center > 2150 Shattuck Ave. | Suite 408 | Berkeley, CA 94704-1345 > T: 510-845-4876 x 14 <tel:510-845-4876%C2%A0x%C2%A014> > <a href=3D"http://starkey.com" target=3D"_blank">starkey.com</a> <<= a href=3D"http://www.starkey.com/" target=3D"_blank">http://www.starkey.com= /</a>> =A0| <a href=3D"http://starkeyresearch.com" target=3D"_blank">sta= rkeyresearch.com</a> > <<a href=3D"http://www.starkeyresearch.com" target=3D"_blank">http:= //www.starkeyresearch.com</a>> =A0| map > <<a href=3D"http://maps.google.com/maps?q=3D3D2150+Shattuck+Avenue,= +Berkeley,+CA+94704&h=3D l=3D3Den&" target=3D"_blank">http://maps.google.com/maps?q=3D3D2150+Sha= ttuck+Avenue,+Berkeley,+CA+94704&h=3D l=3D3Den&</a> > ll=3D3D37.869941,-122.268219&spn=3D3D0.028152,0.038066&sll=3D3= D37.86923,-122.273197&s=3D spn=3D3D0 > .056305,0.076132&z=3D3D15> =A0| email <mailto:<a href=3D"mai= lto:william_woods@xxxxxxxx">william_woods@xxxxxxxx</a>> > =3D20 > The assumptions are: > 1. Talkers have no hearing loss and no cognitive loss. > 2. The talkers are facing each other and speaking in their =3DB3mother= tongue=3D =3DB2. > 3. We know the percent-correct (%C) targeted by the talkers and it is = les=3D s > than 100% (it=3DB9s a challenging environment). > 4. We know the nature of the speech in such a conversation, from a low= -co=3D ntext > vs. high-context perspective. > 5. The talkers are within their critical distance (i.e., ignore reverb= era=3D tion > of talkers=3DB9 speech). > 6. We know the long-term spectral shape of speech at the eardrums. > 7. We know the long-term spectral shape of noise at the eardrums. > 8. The noise is stationary. > 9. The noise is diffuse. > 10. The diffuse noise in combination with diotic direct-wave target sp= eec=3D h > generates the equivalent of an =3DB3internal=3DB2 wideband binaural SN= R improveme=3D nt of > ~1.0 dB over monaural listening. > 11. Overall level is not too high (i.e., no =3DB3roll-over=3DB2 effect= for > intelligibility has occurred). > 12. We=3DB9re not including lip reading. > =3D20 > These assumptions allow one to, first, determine the AI needed to achi= eve=3D =A0the > assumed target %C given the assumed type of speech, and, second, deter= min=3D e the > SNR required with the assumed spectral shapes to obtain that AI. > =3D20 > For instance, assuming talkers want 95%C with =3DB3unfamiliar sentence= s=3DB2 then > using the polynomial fits from Sherbecoe and Studebaker (JASA 1990) of= th=3D e > ANSI S3.5-1969 transfer functions between AI and %C, our talkers would= ne=3D ed an > AI of 0.45. If we assume the noise and speech have the same long-term = spe=3D ctral > shape then the SNR can be determined from (SNR+12)/30=3D3D0.45 (stayin= g with =3D the > 1969 AI method), yielding SNR =3D3D 1.5 dB. =A0Subtracting the binaura= l SNR > improvement yields 0.5 dB. > =3D20 > From: AUDITORY - Research in Auditory Perception > [mailto:<a href=3D"mailto:AUDITORY@xxxxxxxx">AUDITORY@xxxxxxxx= ILL.CA</a>] On Behalf Of Andy Sabin > Sent: Wednesday, January 22, 2014 9:53 AM >=3D20 > To: <a href=3D"mailto:AUDITORY@xxxxxxxx">AUDITORY@xxxxxxxx= CA</a> > Subject: Reference for typical SNRs is public spaces > =3D20 > Hi List,=3D20 > =3D20 > Can anyone point me to a reference showing SNRs that are typically obs= erv=3D ed in > public spaces (e.g., restaurants, bars ...etc)? I can find this info f= or > overall SPL, but am having a hard time finding it for SNR. > =3D20 > Thanks > Andy Sabin > =3D20 >=3D20 >=3D20 --B_<a href=3D"tel:3473483561" value=3D"+13473483561">3473483561</a>_123336= 98 Content-type: text/html; =A0 =A0 =A0 =A0 charset=3D"ISO-8859-1" Content-transfer-encoding: quoted-printable <html><head></head><body style=3D3D"word-wrap: br= eak-word; -webkit-nbsp-mode: s=3D pace; -webkit-line-break: after-white-space; color: rgb(0, 0, 0); font-size= :=3D =A014px; font-family: Calibri, sans-serif;"><div>I agree with= Bill except for =3D one (crucial) assumption: in a cocktail-party situation the noise is not st= a=3D tionary (although, as Jont Allen once suggested, adding talkers to the babb= l=3D e will make it approach stationarity). So, the 0.5 dB SNR is workable in a = b=3D road statistical sense and would have to be adjusted almost on a case-by-ca= s=3D e basis.</div><div> </div><div>-Pierre<= /div><div> </div> SRC_BODY_SECTION"><div style=3D3D"font-family:Calibri; font= -size:11pt; text-alig=3D n:left; color:black; BORDER-BOTTOM: medium none; BORDER-LEFT: medium none; = P=3D ADDING-BOTTOM: 0in; PADDING-LEFT: 0in; PADDING-RIGHT: 0in; BORDER-TOP: #b5c= 4=3D df 1pt solid; BORDER-RIGHT: medium none; PADDING-TOP: 3pt"><span= style=3D3D"font=3D -weight:bold">From: "Richard F. Lyon" &= lt;<a href=3D3D"mailto:<a href=3D"mailto:dicklyon@xxxxxxxx">dicklyon@xxxxxxxx</a>= =3D <a href=3D"http://CM.ORG" target=3D"_blank">CM.ORG</a>"><a href=3D"= mailto:dicklyon@xxxxxxxx">dicklyon@xxxxxxxx</a></a>&gt; &l= t;span style=3D3D"font-weight:bold">Reply-To:=3D =A0 "Richard F. Lyon" &lt;<a href=3D3D"m= ailto:<a href=3D"mailto:dicklyon@xxxxxxxx">dicklyon@xxxxxxxx</a>">dic= klyon@xxxxxxxx=3D <a href=3D"http://M.ORG" target=3D"_blank">M.ORG</a></a>&gt;<b= r>Date: = Friday, Januar=3D y 24, 2014 at 10:16 PM To: &lt;<a =3D href=3D3D"mailto:<a href=3D"mailto:AUDITORY@xxxxxxxx">AUDITORY@xxxxxxxx= LISTS.MCGILL.CA</a>"><a href=3D"mailto:AUDITORY@xxxxxxxx">AU= DITORY@xxxxxxxx</a></a>&gt; <=3D span style=3D3D"font-weight:bold">Subject: Re: [= AUDITORY] Reference for =3D typical SNRs is public spaces </div><div> &lt= ;/div><div dir=3D3D"ltr"><div>I =3D think Bill's point, which I agree with, is that the SNR is not determin= ed by=3D =A0the noise in the space as much as it is by people trying to communicate.= &nb=3D sp; The SNR that he estimates is not "in the space", but rather &= quot;at the ears=3D =A0of the listener" when the talker is trying to communicate to that p= articula=3D r listener, above the noise.&nbsp; Other people trying to communicate t= o dif=3D ferent listeners make noise for this one.&nbsp; The SNR is therefore ro= ughly=3D =A0constant, somewhere near 0 dB, almost anywhere that's not too quiet.= <br=3D ></div>For me, it's a little higher, after I make people speak= up. Di=3D ck </div><div class=3D3D"gmail_extra"= > <div class=3D3D"gmail_quote">On=3D<= br> =A0Fri, Jan 24, 2014 at 1:51 PM, Bill Woods &lt;<a href=3D3D"mailt=3D <a href=3D"mailto:o%3ABill_Woods@xxxxxxxx">o:Bill_Woods@xxxxxxxx</a>&= quot; target=3D3D"_blank"><a href=3D"mailto:Bill_Woods@xxxxxxxx= .com">Bill_Woods@xxxxxxxx</a></a>&gt;</sp=3D an> wrote: <blockquote class=3D3D"gmail_quote" sty= le=3D3D"margin:0 0 0 .8ex;borde=3D r-left:1px #ccc solid;padding-left:1ex"><div><=3D span style=3D3D"font-size:12pt"><div>Dear List,</div&= gt;<div> Roman">&nbsp;</div><div>Since we&#= 8217;re assuming Andy is asking abo=3D ut speech-to-noise ratio when referring to SNR in public spaces, I wondered= =3D if the question couldn&#8217;t be answered by back-of-the-envelope calc= ulati=3D ons using some assumptions and articulation index theory (AI).&nbsp; &l= t;/div><=3D div>&nbsp;</div><div>It turned out there was quite a lon= g list of assumption=3D s behind the calculation (see below), but the resultant simple calculation = y=3D ielded a SNR of 0.5 dB, which is quite in line with the values other poster= s=3D =A0have indicated may actually be present. Two other facts became apparent as I did this, however, and further motivate my= =3D =A0posting. </div><div>&nbsp;</div><div>First, = the long list of assumptions ma=3D kes apparent the long list of influences on the SNR in such a situation. Th= e=3D se influences need to be characterized in any attempt to generalize from SN= R=3D =A0measurements in a given scenario, a fact other posters have alluded to.</div><div>&nbsp;</div><div style=3D3D"mar= gin-bottom:12pt">Second, it is imp=3D ortant to note that much work has been done recently to model or empiricall= y=3D =A0characterize these influences on speech intelligibility, and that, while= th=3D e models may require more computational power than found on the back of an envelope, they are no problem to execute on current lapto= =3D p computers. The implication is that it should not be very difficult to det= e=3D rmine a reasonably-accurate distribution of SNRs over wide variation in the= =3D assumed listening scenario without any measurements. </div><div>Cheers,</div><div>Bill= </div><div> imes New Roman">&nbsp;</div><table style= =3D3D"width:282.75pt;margin-left:=3D 5.4pt" width=3D3D"471"><colgroup><col style=3D3D= "width:282.75pt" width=3D3D"471"></colgr=3D oup><tbody><tr><td>Bill Woods, PhD <fo=3D nt color=3D3D"#595959">Principal Research Scientist=3D &nbsp;|&nbsp;Starkey Hearing Research Center 2150 Shattuck Ave.&nbsp;|&nbsp;Suite 408&nbsp;|&nbsp;Berkel= ey,&nbsp;CA&nbsp=3D ;94704-1345 T:&nbsp;<a href=3D3D"tel:510-845-4876%C2%A0x%C2%A014" valu= e=3D3D"+15108454876" targ=3D et=3D3D"_blank">510-845-4876&nbsp;x&nbsp;14</a> = </td></tr><=3D tr><td><a href= =3D3D"<a href=3D"http://www.starkey.com/" target=3D"_blank">http://www= .starkey.com/</a>" target=3D =3D3D"_blank"><u&gt= ;st=3D <a href=3D"http://arkey.com" target=3D"_blank">arkey.com</a></= span></a> e=3D3D"font-size:10pt">&nbsp;&lt= ;font color=3D3D"#595959" face=3D3D"Arial"><s=3D<= br> pan style=3D3D"font-size:10pt">|<= font color=3D3D"#0000F6" face=3D3D"Arial"=3D >&nbsp;</span&gt= ;<a href=3D3D"<a href=3D"http://www.starke" target=3D"= _blank">http://www.starke</a>=3D <a href=3D"http://yresearch.com" target=3D"_blank">yresearch.com</a>" = target=3D3D"_blank"> font-size:10pt"><a href=3D"http://starkeyresearch.com" tar= get=3D"_blank">starkeyresearch.com</a>&= lt;/a> 00F6" face=3D3D"Arial">&nbsp; lor=3D3D"#595959" face=3D3D"Arial">|<font=3D =A0color=3D3D"#0000F6" face=3D3D"Arial">&nbsp;</fo=3D nt><a href=3D3D"<a href=3D"http://maps.google.com/maps?q=3D3D215= 0+Shattuck+Avenue,+Berkeley,+CA=3D +94704&amp;hl=3D3Den&amp;ll=3D3D37.869941,-122.268219&amp;spn= =3D3D0.028152,0.038066&amp;=3D sll=3D3D37.86923,-122.273197&amp;sspn=3D3D0.056305,0.076132&amp;z= =3D3D15" target=3D"_blank">http://maps.google.com/maps?q=3D3D2150+Shattuck+= Avenue,+Berkeley,+CA=3D +94704&amp;hl=3D3Den&amp;ll=3D3D37.869941,-122.268219&amp;spn= =3D3D0.028152,0.038066&amp;=3D sll=3D3D37.86923,-122.273197&amp;sspn=3D3D0.056305,0.076132&amp;z= =3D3D15</a>" target=3D3D"_blank=3D ">map&lt= ;/s=3D pan></a> t">&nbsp; size:10pt">| nt-size:10pt">&nbsp;<a href=3D3D&q= uot;mailto:<a href=3D"mailto:william_woods@xxxxxxxx">william_woods@xxxxxxxx= ey.com</a>"=3D =A0target=3D3D"_blank"> ">email</a></= font></td></tr><tr><td> ew Roman"></td></tr><tr><td>&l= t;font face=3D3D"Times New Roman"></td>=3D<= br> </tr></tbody></table><div>&nbsp;</div><d=3D iv>The assumptions are:</div><ol style=3D3D"margin:0;paddi= ng-left:36pt"><li>Talk=3D ers have no hearing loss and no cognitive loss.</li><li>The tal= kers are faci=3D ng each other and speaking in their &#8220;mother tongue&#8221;.&lt= ;/li><li>We =3D know the percent-correct (%C) targeted by the talkers and it is less than 1= 0=3D 0% (it&#8217;s a challenging environment).</li><li>We know = the nature of the=3D =A0speech in such a conversation, from a low-context vs. high-context persp= ect=3D ive.</li><li>The talkers are within their critical distance (i.= e., ignore re=3D verberation of talkers&#8217; speech).</li><li> We know the long-term spectral shape of speech at the eardrums.</li>&= lt;li>We =3D know the long-term spectral shape of noise at the eardrums.</li><l= i>The nois=3D e is stationary.</li><li>The noise is diffuse.</li><li= >The diffuse noise in =3D combination with diotic direct-wave target speech generates the equivalent = o=3D f an &#8220;internal&#8221; wideband binaural SNR improvement of ~1= .0 dB ove=3D r monaural listening.</li><li>Overall level is not too high (i.= e., no &#8220=3D ;roll-over&#8221; effect for intelligibility has occurred).</li>&= lt;li>We&#8217=3D ;re not including lip reading.</li></ol><div>&n=3D bsp;</div><div>These assumptions allow one to, fir= st, determine the A=3D I needed to achieve the assumed target %C given the assumed type of speech,= =3D and, second, determine the SNR required with the assumed spectral shapes to= =3D obtain that AI.</div><div>&nbsp;</div><div>For = instance, assuming talkers wa=3D nt 95%C with &#8220;unfamiliar sentences&#8221; then using the poly= nomial fi=3D ts from Sherbecoe and Studebaker (JASA 1990) of the ANSI S3.5-1969 transfer= =3D functions between AI and %C, our talkers would need an AI of 0.45. If we as= s=3D ume the noise and speech have the same long-term spectral shape then the SNR ca= =3D n be determined from (SNR+12)/30=3D3D0.45 (staying with the 1969 AI method)= , yie=3D lding SNR =3D3D 1.5 dB.&nbsp; Subtracting the binaural SNR improvement = yields 0.=3D 5 dB.&nbsp; </div><div>&nbsp;</div><div>=3D <div class=3D3D"im">From:</b=3D<b= r> > AUDITORY - Research in Auditory Perception [<a href=3D3D"mailt= o:<a href=3D"mailto:AUDITORY@xxxxxxxx">AUDITORY@xxxxxxxx</a>=3D .<a href=3D"http://MCGILL.CA" target=3D"_blank">MCGILL.CA</a>" target= =3D3D"_blank">mailto:<a href=3D"mailto:AUDITORY@xxxxxxxx= A">AUDITORY@xxxxxxxx</a></a>] On Behal=3D f Of Andy Sabin </div>Sent: Wed= nesday, January 22, 2014 9:53 A=3D M<div class=3D3D"im"> To: <a= href=3D3D"mailto:<a href=3D"mailto:AUDITORY@xxxxxxxx">AUDITORY= @xxxxxxxx</a>" ta=3D rget=3D3D"_blank"><a href=3D"mailto:AUDITORY@xxxxxxxx">= AUDITORY@xxxxxxxx</a></a> Subject:= Reference for =3D typical SNRs is public spaces</div></div= ><div class=3D3D"im"><div>=3D &nbsp;&lt= ;/div><div> oman">Hi List,&nbsp;</div><div><fon= t face=3D3D"Times New Roman">&nbsp;</f=3D ont></div><div>Can anyone point me to a referen=3D ce showing SNRs that are typically observed in public spaces (e.g., restaur= a=3D nts, bars ...etc)? I can find this info for overall SPL, but am having a ha= r=3D d time finding it for SNR.&nbsp;</div><div>&lt= ;font face=3D3D"Times New Roma=3D n">&nbsp;</div><div>Thanks</div><=3D div>Andy Sabin</font= ></div><div> es New Roman">&nbsp;</div></div></s= pan></div></blockquote></di=3D v> </div></body></html> --B_<a href=3D"tel:3473483561" value=3D"+13473483561">3473483561</a>_123336= 98-- Date: =A0 =A0Sun, 26 Jan 2014 09:54:31 +0000 From: =A0 =A0"Smits, Cas" <<a href=3D"mailto:C.Smits@xxxxxxxx">= C.Smits@xxxxxxxx</a>> Subject: Re: Reference for typical SNRs is public spaces MIME-Version: 1.0 Content-Type: multipart/alternative; boundary=3D"_000_5801ADC15F4A8D4C= AC8F4B3195FA6327EF3B322BSPMXMBX32vumcnl_" --_000_5801ADC15F4A8D4CAC8F4B3195FA6327EF3B322BSPMXMBX32vumcnl_ Content-Type: text/plain; charset=3D"us-ascii" Content-Transfer-Encoding: quoted-printable We used a method to determine the relation between AI/SII and SNR for fluct= =3D uating noise in a way quite similar to the way Bill proposed (Smits and Fes= =3D ten, 2013). We found a different relationship between SII and SNR for two-b= =3D and speech modulated noise compared to steady-state noise (a 2-dB increase = =3D in SNR for fluctuating noise gives almost the same increase in SII as a 1-d= =3D B increase in SNR for steady-state noise). This method works well in lab se= =3D ttings but I think the uncertainties are too large to make reliable predict= =3D ions for real-life situations. The effects of speaking style, speech materi= =3D al, talker etc on percentage correct (the speech recognition function) are = =3D large and the (backward) calculations give large uncertainties in SNR estim= =3D ates. In situations with fluctuating noise the use of the steady-state nois= =3D e relationship between SII and SNR could give large errors. With interferin= =3D g talkers (informational masking) things get even more complicated. But, coming back to the question, normal-hearing listeners can follow most = =3D conversations in public spaces with some difficulty (often in situations wi= =3D th multitalker babble which is very similar to stationairy noise). Thus, SN= =3D Rs around 0-5 dB SNR sounds plausible... Cas Smits Smits C, Festen JM. The interpretation of speech reception threshold data i= =3D n normal-hearing and hearing-impaired listeners: II. Fluctuating noise. J Acoust Soc Am. 2013 May;133(5):3004-15 <a href=3D"http://dr.ir" target=3D"_blank">dr.ir</a>. J.C.M. (Cas) Smits, k= linisch fysicus-audioloog | Universitair Audiol=3D ogisch Centrum | VU medisch centrum Afdeling KNO-Audiologie | EMGO Institute for Health and Care Research Secretariaat Audiologie, Postbus 7057, 1007 MB Amsterdam T (020) 444 09 69 | F (020) 444 09 83 | E c.smits@xxxxxxxx<mailto:<a href= =3D"mailto:c.smits@xxxxxxxx">c.smits@xxxxxxxx</a>=3D <a href=3D"http://mc.nl" target=3D"_blank">mc.nl</a>> | W <a href=3D"htt= p://www.AC-VUmc.nl" target=3D"_blank">www.AC-VUmc.nl</a><<a href=3D"http= ://www.AC-VUmc.nl" target=3D"_blank">http://www.AC-VUmc.nl</a>> =A0<a hr= ef=3D"http://www.CI-VUmc.nl" target=3D"_blank">www.CI-VUmc.nl</a><<a hre= f=3D"http://www" target=3D"_blank">http://www</a>=3D .CI-VUmc.nl> Disclaimer Deze e-mail is vertrouwelijk en alleen bestemd voor de geadresseerde. Als d= =3D eze e-mail niet voor u bestemd is, wordt u verzocht dit aan de afzender te = =3D melden, het bericht niet verder te verspreiden en direct te vernietigen. VU= =3D =A0medisch centrum staat niet in voor de juiste en volledige overbrenging v= an=3D =A0de inhoud van deze e-mail. Van: AUDITORY - Research in Auditory Perception [mailto:<a href=3D"mailto:A= UDITORY@xxxxxxxx">AUDITORY@xxxxxxxx</a>=3D <a href=3D"http://LL.CA" target=3D"_blank">LL.CA</a>] Namens Pierre Divenyi= Verzonden: Saturday, 25 January, 2014 17:33 Aan: <a href=3D"mailto:AUDITORY@xxxxxxxx">AUDITORY@xxxxxxxx</= a> Onderwerp: Re: [AUDITORY] Reference for typical SNRs is public spaces I agree with Bill except for one (crucial) assumption: in a cocktail-party = =3D situation the noise is not stationary (although, as Jont Allen once suggest= =3D ed, adding talkers to the babble will make it approach stationarity). So, t= =3D he 0.5 dB SNR is workable in a broad statistical sense and would have to be= =3D =A0adjusted almost on a case-by-case basis. -Pierre From: "Richard F. Lyon" <<a href=3D"mailto:dicklyon@xxxxxxxx">d= icklyon@xxxxxxxx</a><mailto:<a href=3D"mailto:dicklyon@xxxxxxxx">dicklyon@xxxxxxxx= ACM.ORG</a>>> Reply-To: "Richard F. Lyon" <<a href=3D"mailto:dicklyon@xxxxxxxx= G">dicklyon@xxxxxxxx</a><mailto:<a href=3D"mailto:dicklyon@xxxxxxxx">dickl= yon@xxxxxxxx</a>>> Date: Friday, January 24, 2014 at 10:16 PM To: <<a href=3D"mailto:AUDITORY@xxxxxxxx">AUDITORY@xxxxxxxx= A</a><mailto:<a href=3D"mailto:AUDITORY@xxxxxxxx">AUDITORY@xxxxxxxx= MCGILL.CA</a>>> Subject: Re: [AUDITORY] Reference for typical SNRs is public spaces I think Bill's point, which I agree with, is that the SNR is not determ= ined=3D =A0by the noise in the space as much as it is by people trying to communica= te=3D . =A0The SNR that he estimates is not "in the space", but rather = "at the ears=3D =A0of the listener" when the talker is trying to communicate to that p= articul=3D ar listener, above the noise. =A0Other people trying to communicate to diff= er=3D ent listeners make noise for this one. =A0The SNR is therefore roughly cons= ta=3D nt, somewhere near 0 dB, almost anywhere that's not too quiet. For me, it's a little higher, after I make people speak up. Dick On Fri, Jan 24, 2014 at 1:51 PM, Bill Woods <<a href=3D"mailto:Bill_Wood= s@xxxxxxxx">Bill_Woods@xxxxxxxx</a><mailto:=3D <a href=3D"mailto:Bill_Woods@xxxxxxxx">Bill_Woods@xxxxxxxx</a>>&gt= ; wrote: Dear List, Since we're assuming Andy is asking about speech-to-noise ratio when re= ferr=3D ing to SNR in public spaces, I wondered if the question couldn't be ans= were=3D d by back-of-the-envelope calculations using some assumptions and articulat= =3D ion index theory (AI). It turned out there was quite a long list of assumptions behind the calcula= =3D tion (see below), but the resultant simple calculation yielded a SNR of 0.5= =3D =A0dB, which is quite in line with the values other posters have indicated = ma=3D y actually be present. Two other facts became apparent as I did this, howev= =3D er, and further motivate my posting. First, the long list of assumptions makes apparent the long list of influen= =3D ces on the SNR in such a situation. These influences need to be characteriz= =3D ed in any attempt to generalize from SNR measurements in a given scenario, = =3D a fact other posters have alluded to. Second, it is important to note that much work has been done recently to mo= =3D del or empirically characterize these influences on speech intelligibility,= =3D =A0and that, while the models may require more computational power than fou= nd=3D =A0on the back of an envelope, they are no problem to execute on current la= pt=3D op computers. The implication is that it should not be very difficult to de= =3D termine a reasonably-accurate distribution of SNRs over wide variation in t= =3D he assumed listening scenario without any measurements. Cheers, Bill Bill Woods, PhD Principal Research Scientist | Starkey Hearing Research Center 2150 Shattuck Ave. | Suite 408 | Berkeley, CA 94704-1345 T: <a href=3D"tel:510-845-4876%20x%2014" value=3D"+15108454876">510-845-487= 6 x 14</a><tel:510-845-4876%C2%A0x%C2%A014> <a href=3D"http://starkey.com" target=3D"_blank">starkey.com</a><<a href= =3D"http://www.starkey.com/" target=3D"_blank">http://www.starkey.com/</a>&= gt; | <a href=3D"http://starkeyresearch.com" target=3D"_blank">starkeyresea= rch.com</a><<a href=3D"http://www.stark" target=3D"_blank">http://www.st= ark</a>=3D <a href=3D"http://eyresearch.com" target=3D"_blank">eyresearch.com</a>> = | map<<a href=3D"http://maps.google.com/maps?q=3D3D2150+Shattuck+Avenue,= =3D +Berkeley,+CA+94704&hl=3D3Den&ll=3D3D37.869941,-122.268219&spn= =3D3D0.028152,0.038=3D 066&sll=3D3D37.86923,-122.273197&sspn=3D3D0.056305,0.076132&z= =3D3D15" target=3D"_blank">http://maps.google.com/maps?q=3D3D2150+Shattuck+= Avenue,=3D +Berkeley,+CA+94704&hl=3D3Den&ll=3D3D37.869941,-122.268219&spn= =3D3D0.028152,0.038=3D 066&sll=3D3D37.86923,-122.273197&sspn=3D3D0.056305,0.076132&z= =3D3D15</a>> | email<mai=3D <a href=3D"mailto:lto%3Awilliam_woods@xxxxxxxx">lto:william_woods@xxxxxxxx= y.com</a>> The assumptions are: 1. =A0 =A0 =A0Talkers have no hearing loss and no cognitive loss. 2. =A0 =A0 =A0The talkers are facing each other and speaking in their &quot= ;mother ton=3D gue". 3. =A0 =A0 =A0We know the percent-correct (%C) targeted by the talkers and = it is =3D less than 100% (it's a challenging environment). 4. =A0 =A0 =A0We know the nature of the speech in such a conversation, from= a low=3D -context vs. high-context perspective. 5. =A0 =A0 =A0The talkers are within their critical distance (i.e., ignore = reverb=3D eration of talkers' speech). 6. =A0 =A0 =A0We know the long-term spectral shape of speech at the eardrum= s. 7. =A0 =A0 =A0We know the long-term spectral shape of noise at the eardrums= . 8. =A0 =A0 =A0The noise is stationary. 9. =A0 =A0 =A0The noise is diffuse. 10. =A0The diffuse noise in combination with diotic direct-wave target spee= ch=3D =A0generates the equivalent of an "internal" wideband binaural SN= R improvemen=3D t of ~1.0 dB over monaural listening. 11. =A0Overall level is not too high (i.e., no "roll-over" effect= for intelli=3D gibility has occurred). 12. =A0We're not including lip reading. These assumptions allow one to, first, determine the AI needed to achieve t= =3D he assumed target %C given the assumed type of speech, and, second, determi= =3D ne the SNR required with the assumed spectral shapes to obtain that AI. For instance, assuming talkers want 95%C with "unfamiliar sentences&qu= ot; then u=3D sing the polynomial fits from Sherbecoe and Studebaker (JASA 1990) of the A= =3D NSI S3.5-1969 transfer functions between AI and %C, our talkers would need = =3D an AI of 0.45. If we assume the noise and speech have the same long-term sp= =3D ectral shape then the SNR can be determined from (SNR+12)/30=3D3D0.45 (stay= in=3D g with the 1969 AI method), yielding SNR =3D3D 1.5 dB. =A0Subtracting the b= inau=3D ral SNR improvement yields 0.5 dB. From: AUDITORY - Research in Auditory Perception [mailto:<a href=3D"mailto:= AUDITORY@xxxxxxxx">AUDITORY@xxxxxxxx</a>=3D <a href=3D"http://ILL.CA" target=3D"_blank">ILL.CA</a>] On Behalf Of Andy S= abin Sent: Wednesday, January 22, 2014 9:53 AM To: <a href=3D"mailto:AUDITORY@xxxxxxxx">AUDITORY@xxxxxxxx</a= ><mailto:<a href=3D"mailto:AUDITORY@xxxxxxxx">AUDITORY@xxxxxxxx= LL.CA</a>> Subject: Reference for typical SNRs is public spaces Hi List, Can anyone point me to a reference showing SNRs that are typically observed= =3D =A0in public spaces (e.g., restaurants, bars ...etc)? I can find this info = fo=3D r overall SPL, but am having a hard time finding it for SNR. Thanks Andy Sabin --_000_5801ADC15F4A8D4CAC8F4B3195FA6327EF3B322BSPMXMBX32vumcnl_ Content-Type: text/html; charset=3D"us-ascii" Content-Transfer-Encoding: quoted-printable <html xmlns:v=3D3D"urn:schemas-microsoft-com:vml" xmlns:o=3D3D= "urn:schemas-micr=3D osoft-com:office:office" xmlns:w=3D3D"urn:schemas-microsoft-com:o= ffice:word" =3D xmlns:x=3D3D"urn:schemas-microsoft-com:office:excel" xmlns:p=3D3D= "urn:schemas-m=3D icrosoft-com:office:powerpoint" xmlns:a=3D3D"urn:schemas-microsof= t-com:office=3D :access" xmlns:dt=3D3D"uuid:C2F41010-65B3-11d1-A29F-00AA00C14882&= quot; xmlns:s=3D3D"=3D uuid:BDC6E3F0-6DA3-11d1-A2A3-00AA00C14882" xmlns:rs=3D3D"urn:sche= mas-microsof=3D t-com:rowset" xmlns:z=3D3D"#RowsetSchema" xmlns:b=3D3D"= urn:schemas-microsoft-co=3D m:office:publisher" xmlns:ss=3D3D"urn:schemas-microsoft-com:offic= e:spreadshee=3D t" xmlns:c=3D3D"urn:schemas-microsoft-com:office:component:spread= sheet" xmlns=3D :odc=3D3D"urn:schemas-microsoft-com:office:odc" xmlns:oa=3D3D&quo= t;urn:schemas-micro=3D soft-com:office:activation" xmlns:html=3D3D"<a href=3D"http://www= .w3.org/TR/REC-html40" target=3D"_blank">http://www.w3.org/TR/REC-html40</a= >" =3D xmlns:q=3D3D"<a href=3D"http://schemas.xmlsoap.org/soap/envelope/" tar= get=3D"_blank">http://schemas.xmlsoap.org/soap/envelope/</a>" xmlns:rt= c=3D3D"<a href=3D"http://m" target=3D"_blank">http://m</a>=3D <a href=3D"http://icrosoft.com/officenet/conferencing" target=3D"_blank">ic= rosoft.com/officenet/conferencing</a>" xmlns:D=3D3D"DAV:" xm= lns:Repl=3D3D"http://=3D <a href=3D"http://schemas.microsoft.com/repl/" target=3D"_blank">schemas.mi= crosoft.com/repl/</a>" xmlns:mt=3D3D"<a href=3D"http://schemas.mi= crosoft.com/share=3D point/soap/meetings/" target=3D"_blank">http://schemas.microsoft.com/share= =3D point/soap/meetings/</a>" xmlns:x2=3D3D"<a href=3D"http://schemas= .microsoft.com/office/excel=3D /2003/xml" target=3D"_blank">http://schemas.microsoft.com/office/excel=3D<b= r> /2003/xml</a>" xmlns:ppda=3D3D"<a href=3D"http://www.passport.com= /NameSpace.xsd" target=3D"_blank">http://www.passport.com/NameSpace.xsd</a>= " xmlns:ois=3D =3D3D"<a href=3D"http://schemas.microsoft.com/sharepoint/soap/ois/" ta= rget=3D"_blank">http://schemas.microsoft.com/sharepoint/soap/ois/</a>"= xmlns:dir=3D3D"http://=3D <a href=3D"http://schemas.microsoft.com/sharepoint/soap/directory/" target= =3D"_blank">schemas.microsoft.com/sharepoint/soap/directory/</a>" xmln= s:ds=3D3D"<a href=3D"http://www.w3" target=3D"_blank">http://www.w3</a= >=3D .org/2000/09/xmldsig#" xmlns:dsp=3D3D"<a href=3D"http://schemas.m= icrosoft.com/sharepoint=3D" target=3D"_blank">http://schemas.microsoft.com/= sharepoint=3D</a> /dsp" xmlns:udc=3D3D"<a href=3D"http://schemas.microsoft.com/data= /udc" target=3D"_blank">http://schemas.microsoft.com/data/udc</a>" xml= ns:xsd=3D3D"http=3D ://<a href=3D"http://www.w3.org/2001/XMLSchema" target=3D"_blank">www.w3.or= g/2001/XMLSchema</a>" xmlns:sub=3D3D"<a href=3D"http://schemas.mi= crosoft.com/sha=3D repoint/soap/2002/1/alerts/" target=3D"_blank">http://schemas.microsoft.com= /sha=3D repoint/soap/2002/1/alerts/</a>" xmlns:ec=3D3D"<a href=3D"http://= www.w3.org/2001/04/xmlenc#" target=3D"_blank">http://www.w3.org/2001/04/xml= enc#</a>"=3D =A0xmlns:sp=3D3D"<a href=3D"http://schemas.microsoft.com/sharepoint/" = target=3D"_blank">http://schemas.microsoft.com/sharepoint/</a>" xmlns:= sps=3D3D"http://=3D <a href=3D"http://schemas.microsoft.com/sharepoint/soap/" target=3D"_blank"= >schemas.microsoft.com/sharepoint/soap/</a>" xmlns:xsi=3D3D"<a hr= ef=3D"http://www.w3.org/2001=3D /XMLSchema-instance" target=3D"_blank">http://www.w3.org/2001=3D /XMLSchema-instance</a>" xmlns:udcs=3D3D"<a href=3D"http://schema= s.microsoft.com/data/udc/so=3D" target=3D"_blank">http://schemas.microsoft.= com/data/udc/so=3D</a> ap" xmlns:udcxf=3D3D"<a href=3D"http://schemas.microsoft.com/data= /udc/xmlfile" target=3D"_blank">http://schemas.microsoft.com/data/udc/xmlfi= le</a>" xmlns:udc=3D p2p=3D3D"<a href=3D"http://schemas.microsoft.com/data/udc/parttopart" = target=3D"_blank">http://schemas.microsoft.com/data/udc/parttopart</a>&quot= ; xmlns:m=3D3D"http://=3D <a href=3D"http://schemas.microsoft.com/office/2004/12/omml" target=3D"_bla= nk">schemas.microsoft.com/office/2004/12/omml</a>" xmlns:st=3D3D"= &#1;" xmlns=3D3D"http:=3D //<a href=3D"http://www.w3.org/TR/REC-html40" target=3D"_blank">www.w3.org/= TR/REC-html40</a>"> <head> <meta http-equiv=3D3D"Content-Type" content=3D3D"text/htm= l; charset=3D3Dus-ascii"=3D > <meta name=3D3D"Generator" content=3D3D"Microsoft Word 12= (filtered medium)"> <style>  </style> </head> <body lang=3D3D"EN-US" link=3D3D"blue" vlink=3D3D&qu= ot;purple" style=3D3D"word-wrap: bre=3D ak-word; -webkit-nbsp-mode: space;-webkit-line-break: after-white-space"><br= > <div class=3D3D"Section1"> libri&quot;,&quot;sans-serif&quot;; color:#1F497D">We used a method to determine the relation between A= I/SII an=3D d SNR for fluctuating noise in a way quite similar to the way Bill proposed= =3D =A0(Smits and Festen, =A02013). We found a different relationship between SII and SNR for two-ban= d =3D speech modulated noise compared to steady-state noise (a 2-dB increase in S= =3D NR for fluctuating noise gives almost the same increase in SII as a 1-dB in= =3D crease in SNR for steady-state noise). =A0This method works well in lab settings but I think the uncertainties are= t=3D oo large to make reliable predictions for real-life situations. The effects= =3D =A0of speaking style, speech material, talker etc on percentage correct (th= e =3D speech recognition function) are =A0large and the (backward) calculations give large uncertainties in SNR es= ti=3D mates. In situations with fluctuating noise the use of the steady-state noi= =3D se relationship between SII and SNR could give large errors. With interferi= =3D ng talkers (informational masking) =A0things get even more complicated.<o:p></o:p>&lt= ;/p> libri&quot;,&quot;sans-serif&quot;; color:#1F497D"><o:p>&nbsp;</o:p></p= > libri&quot;,&quot;sans-serif&quot;; color:#1F497D">But, coming back to the question, normal-hearing lis= teners c=3D an follow most conversations in public spaces with some difficulty (often i= =3D n situations with multitalker =A0babble which is very similar to stationairy noise). Thus, SNRs around 0-= 5 =3D dB SNR sounds plausible...<o:p></o:p><br= > libri&quot;,&quot;sans-serif&quot;; color:#1F497D"><o:p>&nbsp;</o:p></p= > libri&quot;,&quot;sans-serif&quot;; color:#1F497D">Cas Smits<o:p></o:p></p&= gt; libri&quot;,&quot;sans-serif&quot;; color:#1F497D"><o:p>&nbsp;</o:p></p= > libri&quot;,&quot;sans-serif&quot;; color:#1F497D"><o:p>&nbsp;</o:p></p= > libri&quot;,&quot;sans-serif&quot;; color:#1F497D">Smits C, Festen JM. The interpretation of speech rec= eption t=3D hreshold data in normal-hearing and hearing-impaired listeners: II. Fluctua= =3D ting noise.<o:p></o:p> libri&quot;,&quot;sans-serif&quot;; color:#1F497D">J Acoust Soc Am. 2013 May;133(5):3004-15<o:p>&= lt;/o:p></=3D p> libri&quot;,&quot;sans-serif&quot;; color:#1F497D"><o:p>&nbsp;</o:p></p= > libri&quot;,&quot;sans-serif&quot;; color:#1F497D"><o:p>&nbsp;</o:p></p= > ctly"> ot;sans-serif&quot;;color:#10137D"><a href=3D"http://dr.ir" tar= get=3D"_blank">dr.ir</a>. J.C.M. (Cas) Smits, klinisch fysi=3D cus-audioloog olor:#10137D">&nbsp; y:&quot;Verdana&quot;,&quot;sans-serif&quot;;color:#10137D&= quot;>| =A0Universitair Audiologisch Centrum | VU medisch centrum</= b> le=3D3D"font-size: 8.0pt;font-family:&quot;Verdana&quot;,&quot;sans-serif&quot= ;;color:black"> ;sans-serif&quot;; color:black">Afdeling KNO-Audiologie | EMGO Institute for Health an= d Care R=3D esearch<o:p&g= t;</o:p> ctly"> quot;,&quot;sans-serif&quot;;color:black">Secretariaat Audi= ologie, Postbus =3D 7057, 1007 MB Amsterdam ana&quot;,&quot;sans-serif&quot;; color:#E5053A">T family: &quot;Verdana&quot;,&quot;sans-serif&quot;;color:black&quot= ;> (020) 444 09 69 |&=3D nbsp; ;Verdana&quot;,&quot;sans-serif&quot;;color:#E5053A">F&= lt;/span> =3D3D"NL" style=3D3D"font-size:8.0pt;font-family:&quot;V= erdana&quot;,&quot;sans=3D -serif&quot;;color:black"> =A0(020) 444 09 83 |&nbsp; t; font-family:&quot;Verdana&quot;,&quot;sans-serif&quot;;colo= r:#E5053A">E</sp=3D an> &quot;,&quot;sans-serif&quot;;color:black">&nbsp;&l= t;/span> r:#10137D"><a href=3D3D"mailto:<a href=3D"mailto:c.smits@xxxxxxxx= mc.nl">c.smits@xxxxxxxx</a>"> nt-size:8.0pt;font-family:&quot;Verdana&quot;,&quot;sans-serif&= amp;quot;">c.smi=3D ts@xxxxxxxx</a> -family:&quot;Verdana&quot;,&quot;sans-serif&quot;;color:bl= ack">&nbsp;| ana&quot;,&quot;sans-serif&quot;; color:#E5053A">W family: &quot;Verdana&quot;,&quot;sans-serif&quot;;color:black&quot= ;> <a href=3D3= D"<a href=3D"http://www.AC-VUmc.nl" target=3D"_blank">http://www.AC-VU= mc.nl</a>"><spa=3D n lang=3D3D"NL" style=3D3D"font-size:8.0pt;font-family: &quot;Verdana&quot;,&quot;sans-serif&quot;"><a href= =3D"http://www.AC-VUmc.nl" target=3D"_blank">www.AC-VUmc.nl</a></span&gt= ;</a></span=3D > ,&quot;sans-serif&quot;;color:black">&nbsp;</span&gt= ; =3D3D"font-size:8.0pt;font-family:&quot;Verdana&quot;,&quo= t;sans-serif&quot;;=3D color:#E5053A">&nbsp; t;Verdana&quot;,&quot;sans-serif&quot;;color:purple">&l= t;a href=3D3D"<a href=3D"http://www." target=3D"_blank">http://www.</a= >=3D CI-VUmc.nl"><a href=3D"http://w= ww.CI-VUmc.nl" target=3D"_blank">www.CI-VUmc.nl</a></a>&= lt;/span> "NL" style=3D3D"color:#10137D"><o:p></o:p&g= t; ctly">&nbsp;<o:p></o:p></p=3D > &nbsp;<o:p=3D ></o:p> gray">Disclaimer<o:p></o:p></= p> y">Deze e-mail is vertrouwelijk en alleen bestemd voor de geadresse= erde. Al=3D s deze e-mail niet voor u bestemd is, wordt u verzocht dit aan de afzender = =3D te melden, het bericht niet verder te =A0verspreiden en direct te vernietigen. VU medisch centrum staat niet in v= oo=3D r de juiste en volledige overbrenging van de inhoud van deze e-mail.</sp= an>=3D <o:p></o:p>&lt= ;/p> ily:&quot;Calibri&quot;,&quot;sans-serif&quot;; color:#1F497D"><o:p>&nbsp;</o:p></p= > ily:&quot;Calibri&quot;,&quot;sans-serif&quot;; color:#1F497D"><o:p>&nbsp;</o:p></p= > ily:&quot;Calibri&quot;,&quot;sans-serif&quot;; color:#1F497D"><o:p>&nbsp;</o:p></p= > <div> <div style=3D3D"border:none;border-top:solid #B5C4DF 1.0pt;padding:= 3.0pt 0cm =3D 0cm 0cm"> family:&quot;Tahoma&quot;,&quot;sans-serif&quot;">V= an: =3D3D"NL" style=3D3D"font-size:10.0pt;font-family:&quot;= Tahoma&quot;,&quot;sans=3D -serif&quot;"> AUDITORY - Research in Auditory Perception [mail= to:<a href=3D"mailto:AUDITORY@xxxxxxxx">AUDITORY@xxxxxxxx</a>=3D <a href=3D"http://LISTS.MCGILL.CA" target=3D"_blank">LISTS.MCGILL.CA</a>]<b= r> Namens Pierre Divenyi Verzonden: Saturday, 25 January, 2014 17:33 <br= > Aan: <a href=3D"mailto:AUDITORY@xxxxxxxx">AUDITOR= Y@xxxxxxxx</a> Onderwerp: Re: [AUDITORY] Reference for typical SNRs is = public space=3D s<o:p></o:p> </div> </div> <o:p>&nbsp;</o:p>&= lt;/p> <div> libri&quot;,&quot;sans-serif&quot;; color:black">I agree with Bill except for one (crucial) assumption:= in a co=3D cktail-party situation the noise is not stationary (although, as Jont Allen= =3D =A0once suggested, adding =A0talkers to the babble will make it approach stationarity). So, the 0.5 d= B =3D SNR is workable in a broad statistical sense and would have to be adjusted = =3D almost on a case-by-case basis.<o:p></o:p></p&g= t; </div> <div> libri&quot;,&quot;sans-serif&quot;; color:black"><o:p>&nbsp;</o:p></p&g= t; </div> <div> libri&quot;,&quot;sans-serif&quot;; color:black">-Pierre<o:p></o:p><= br> </div> <div> libri&quot;,&quot;sans-serif&quot;; color:black"><o:p>&nbsp;</o:p></p&g= t; </div> <div style=3D3D"border:none;border-top:solid #B5C4DF 1.0pt;padding:= 3.0pt 0cm =3D 0cm 0cm"> ;Calibri&quot;,&quot;sans-serif&quot;; color:black">From: &quot;sans-serif&quot;; color:black">&quot;Richard F. Lyon&quot; &lt;<a href= =3D3D"mailto:<a href=3D"mailto:dicklyon@xxxxxxxx">dicklyon@xxxxxxxx</a>=3D .ORG"><a href=3D"mailto:dicklyon@xxxxxxxx">dicklyon@xxxxxxxx</a></= a>&gt; Reply-To: &quot;Richard F. Lyon&quot; &lt;&l= t;a href=3D3D"mailto:<a href=3D"mailto:dicklyon">dicklyon</a>=3D @xxxxxxxx<a href=3D"http://ACM.ORG" target=3D"_blank">ACM.ORG</a>"><a href= =3D"mailto:dicklyon@xxxxxxxx">dicklyon@xxxxxxxx</a></a>&gt;<br&g= t; Date: Friday, January 24, 2014 at 10:16 PM To: &lt;<a href=3D3D"mailto:<a href=3D"mailt= o:AUDITORY@xxxxxxxx">AUDITORY@xxxxxxxx</a>">AUDITORY@xxxxxxxx= LISTS.M=3D <a href=3D"http://CGILL.CA" target=3D"_blank">CGILL.CA</a></a>&gt= ; Subject: Re: [AUDITORY] Reference for typical SNRs is pu= blic spaces<=3D o:p></o:p> </div> <div> libri&quot;,&quot;sans-serif&quot;; color:black"><o:p>&nbsp;</o:p></p&g= t; </div> <div> <div> ize:10.5pt; font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;;colo= r:black">I think=3D =A0Bill's point, which I agree with, is that the SNR is not determined = by the=3D =A0noise in the space as much as it is by people =A0trying to communicate.&nbsp; The SNR that he estimates is not &q= uot;in the=3D =A0space&quot;, but rather &quot;at the ears of the listener&qu= ot; when the t=3D alker is trying to communicate to that particular listener, above the noise= =3D .&nbsp; Other people trying to communicate to different listeners =A0make noise for this one.&nbsp; The SNR is therefore roughly constant= , some=3D where near 0 dB, almost anywhere that's not too quiet.<o:p></o= :p></p=3D > </div> ize:10.5pt; font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;;colo= r:black">For me,=3D =A0it's a little higher, after I make people speak up. Dick<o:p></o:p> </div> <div> ize:10.5pt; font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;;colo= r:black"><o:p>&n=3D bsp;</o:p> <div> libri&quot;,&quot;sans-serif&quot;; color:black">On Fri, Jan 24, 2014 at 1:51 PM, Bill Woods &lt;&l= t;a href=3D3D"mai=3D <a href=3D"mailto:lto%3ABill_Woods@xxxxxxxx">lto:Bill_Woods@xxxxxxxx<= /a>" target=3D3D"_blank"><a href=3D"mailto:Bill_Woods@xxxxxxxx= rkey.com">Bill_Woods@xxxxxxxx</a></a>&gt=3D ; wrote:<o:p></o:p> <div> <div> ;sans-serif&quot;;color:black">Dear List,<o:p></o:p&gt= ; </div> <div> &nbsp; =3D3D"font-family:&quot;Calibri&quot;,&quot;sans-serif&amp= ;quot;;color:black"><o:=3D p></o:p> </div> <div> ;sans-serif&quot;;color:black">Since we&#8217;re assuming A= ndy is asking ab=3D out speech-to-noise ratio when referring to SNR in public spaces, I wondere= =3D d if the question couldn&#8217;t be answered by back-of-the-envelope<br= > =A0calculations using some assumptions and articulation index theory (AI).&= amp;nb=3D sp; <o:p></o:p> </div> <div> ;sans-serif&quot;;color:black">&nbsp;<o:p></o:p&gt= ; </div> <div> ;sans-serif&quot;;color:black">It turned out there was quite a = long list of=3D =A0assumptions behind the calculation (see below), but the resultant simple= c=3D alculation yielded a SNR of 0.5 dB, which is quite in line =A0with the values other posters have indicated may actually be present. Tw= o =3D other facts became apparent as I did this, however, and further motivate my= =3D =A0posting. <o:p></o:p> </div> <div> ;sans-serif&quot;;color:black">&nbsp;<o:p></o:p&gt= ; </div> <div> ;sans-serif&quot;;color:black">First, the long list of assumpti= ons makes ap=3D parent the long list of influences on the SNR in such a situation. These in= =3D fluences need to be characterized in any attempt to generalize =A0from SNR measurements in a given scenario, a fact other posters have all= ud=3D ed to.<o:p></o:p> </div> <div> ;sans-serif&quot;;color:black">&nbsp;<o:p></o:p&gt= ; </div> <div style=3D3D"margin-bottom:12.0pt"> ;sans-serif&quot;;color:black">Second, it is important to note = that much wo=3D rk has been done recently to model or empirically characterize these influe= =3D nces on speech intelligibility, and that, while the models =A0may require more computational power than found on the back of an envelo= pe=3D , they are no problem to execute on current laptop computers. The implicati= =3D on is that it should not be very difficult to determine a reasonably-accura= =3D te distribution of SNRs over wide =A0variation in the assumed listening scenario without any measurements. &l= t;o:p=3D ></o:p> </div> <div> ;sans-serif&quot;;color:black">Cheers,<o:p></o:p>&l= t;/span> </div> <div> ;sans-serif&quot;;color:black">Bill<o:p></o:p></= span> </div> <div> &nbsp; =3D3D"font-family:&quot;Calibri&quot;,&quot;sans-serif&amp= ;quot;;color:black"><o:=3D p></o:p> </div> <table class=3D3D"MsoNormalTable" border=3D3D"0" cel= lpadding=3D3D"0" width=3D3D"377=3D " style=3D3D"width:282.75pt; =A0margin-left:5.4pt"> <tbody> <tr> <td style=3D3D"padding:.75pt .75pt .75pt .75pt"> ans-serif&quot;;color:#17365D">Bill Woods, PhD sans-serif&quot;; =A0 color:#595959">Principal Research Scientist&nbsp;|&nbsp= ;Starkey Hearing R=3D esearch Center 2150 Shattuck Ave.&nbsp;|&nbsp;Suite 408&nbsp;|&nbsp;Berkel= ey,&nbsp;CA&nbsp=3D ;94704-1345 T:&nbsp;<a href=3D3D"tel:510-845-4876%C2%A0x%C2%A014" targ= et=3D3D"_blank">510-8=3D 45-4876&nbsp;x&nbsp;14</a> <o:p></o:p> </td> </tr> <tr> <td style=3D3D"padding:.75pt .75pt .75pt .75pt"> <a href=3D3D"<a href=3D"ht= tp://www.starkey.com/" target=3D"_blank">http://www.starkey.com/</a>" = target=3D3D"_blank=3D "> serif&quot;"><a href=3D"http://starkey.com" target=3D"_blank">s= tarkey.com</a></a> ily:&quot;Arial&quot;,&quot;sans-serif&quot;;color:#0000F6&= quot;>&nbsp;<s=3D pan style=3D3D"font-size:10.0pt;font-family:&quot;Arial&quot;,= &quot;sans-seri=3D f&quot;;color:#595959">| &quot;Arial&quot;,&quot;sans-serif&quot;;color:#0000F6&quot= ;>&nbsp;<a hre=3D f=3D3D"<a href=3D"http://www.starkeyresearch.com" target=3D"_blank">ht= tp://www.starkeyresearch.com</a>" target=3D3D"_blank">&lt= ;span style=3D3D"font-=3D size: =A0 10.0pt;font-family:&quot;Arial&quot;,&quot;sans-serif&q= uot;">starkeyresea=3D <a href=3D"http://rch.com" target=3D"_blank">rch.com</a></a= > uot;,&quot;sans-serif&quot;;color:#0000F6">&nbsp;</s= pan> -size:10.0pt;font-family:&quot;Arial&quot;,&quot;sans-serif&amp= ;quot;;color:#59=3D 5959">| ,&quot;sans-serif&quot;;color:#0000F6">&nbsp;</span&= gt;<a href=3D3D"<a href=3D"http://maps." target=3D"_blank">http://m= aps.</a>=3D <a href=3D"http://google.com/maps?q=3D3D2150&#43;Shattuck&#43;Avenu= e,&#43;Berkeley,&#43;CA&#43;=3D 94704&amp;hl=3D3Den&amp;ll=3D3D37.869941,-122.268219&amp;spn=3D= 3D0.028152,0.03806=3D 6&amp;sll=3D3D37.86923,-122.273197&amp;sspn=3D3D0.056305,0.076132&a= mp;amp;z=3D3D15" target=3D"_blank">google.com/maps?q=3D3D2150&#43;Shatt= uck&#43;Avenue,&#43;Berkeley,&#43;CA&#43;=3D 94704&amp;hl=3D3Den&amp;ll=3D3D37.869941,-122.268219&amp;spn=3D= 3D0.028152,0.03806=3D 6&amp;sll=3D3D37.86923,-122.273197&amp;sspn=3D3D0.056305,0.076132&a= mp;amp;z=3D3D15</a>" t=3D arget=3D3D"_blank"> ot;,&quot;sans-serif&quot;">map</a><s= pan style=3D3D"font-size:10.0pt;f=3D ont-family:&quot;Arial&quot;,&quot;sans-serif&quot;;color:#= 0000F6">&nbsp;</=3D span> ns-serif&quot;;color:#595959">| family:&quot;Arial&quot;,&quot;sans-serif&quot;;color:#0000= F6">&nbsp;</span=3D ><a href=3D3D"mailto:<a href=3D"mailto:william_woods@xxxxxxxx= ">william_woods@xxxxxxxx</a>" target=3D3D"_blank"><= span style=3D =3D3D"font-size:10.0pt;font-family:&quot;Arial&quot;,&quot= ;sans-serif&quot;">=3D email</a><o:p></o:p> </td> </tr> <tr> <td style=3D3D"padding:.75pt .75pt .75pt .75pt"></td>= </tr> <tr> <td style=3D3D"padding:.75pt .75pt .75pt .75pt"></td>= </tr> </tbody> </table> <div> &nbsp; =3D3D"font-family:&quot;Calibri&quot;,&quot;sans-serif&amp= ;quot;;color:black"><o:=3D p></o:p> </div> <div> ;sans-serif&quot;;color:black">The assumptions are:<o:p>&= lt;/o:p> </div> lt:auto; margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> <![if !supportLists]> ans-serif&quot;;color:black"> Ignore">1.&nbsp;&nbs=3D p;&nbsp;&nbsp;&nbsp; <![endif]> t;,&quot;sans-serif&quot;; color:black">Talkers have no hearing loss and no cognitive loss.&lt= ;o:p></o:p>=3D lt:auto; margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> <![if !supportLists]> ans-serif&quot;;color:black"> Ignore">2.&nbsp;&nbs=3D p;&nbsp;&nbsp;&nbsp; <![endif]> t;,&quot;sans-serif&quot;; color:black">The talkers are facing each other and speaking in thei= r &#8220=3D ;mother tongue&#8221;.<o:p></o:p><br= > lt:auto; margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> <![if !supportLists]> ans-serif&quot;;color:black"> Ignore">3.&nbsp;&nbs=3D p;&nbsp;&nbsp;&nbsp; <![endif]> t;,&quot;sans-serif&quot;; color:black">We know the percent-correct (%C) targeted by the talke= rs and i=3D t is less than 100% (it&#8217;s a challenging environment).<o:p>&= lt;/o:p></spa=3D n> lt:auto; margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> <![if !supportLists]> ans-serif&quot;;color:black"> Ignore">4.&nbsp;&nbs=3D p;&nbsp;&nbsp;&nbsp; <![endif]> t;,&quot;sans-serif&quot;; color:black">We know the nature of the speech in such a conversatio= n, from =3D a low-context vs. high-context perspective.<o:p></o:p></span= > lt:auto; margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> <![if !supportLists]> ans-serif&quot;;color:black"> Ignore">5.&nbsp;&nbs=3D p;&nbsp;&nbsp;&nbsp; <![endif]> t;,&quot;sans-serif&quot;; color:black">The talkers are within their critical distance (i.e., = ignore r=3D everberation of talkers&#8217; speech).<o:p></o:p></span= > lt:auto; margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> <![if !supportLists]> ans-serif&quot;;color:black"> Ignore">6.&nbsp;&nbs=3D p;&nbsp;&nbsp;&nbsp; <![endif]> t;,&quot;sans-serif&quot;; color:black">We know the long-term spectral shape of speech at the = eardrums=3D .<o:p></o:p> lt:auto; margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> <![if !supportLists]> ans-serif&quot;;color:black"> Ignore">7.&nbsp;&nbs=3D p;&nbsp;&nbsp;&nbsp; <![endif]> t;,&quot;sans-serif&quot;; color:black">We know the long-term spectral shape of noise at the e= ardrums.=3D <o:p></o:p> lt:auto; margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> <![if !supportLists]> ans-serif&quot;;color:black"> Ignore">8.&nbsp;&nbs=3D p;&nbsp;&nbsp;&nbsp; <![endif]> t;,&quot;sans-serif&quot;; color:black">The noise is stationary.<o:p></o:p></sp= an> lt:auto; margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> <![if !supportLists]> ans-serif&quot;;color:black"> Ignore">9.&nbsp;&nbs=3D p;&nbsp;&nbsp;&nbsp; <![endif]> t;,&quot;sans-serif&quot;; color:black">The noise is diffuse.<o:p></o:p></span&= gt; lt:auto; margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> <![if !supportLists]> ans-serif&quot;;color:black"> Ignore">10.&nbsp; <![endif]> t;,&quot;sans-serif&quot;;color:black">The diffuse noise in= combination wit=3D h diotic direct-wave target speech generates the equivalent of an &#822= 0;in=3D ternal&#8221; wideband binaural SNR improvement of ~1.0 dB over monaura= l =A0listening.<o:p></o:p> lt:auto; margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> <![if !supportLists]> ans-serif&quot;;color:black"> Ignore">11.&nbsp; <![endif]> t;,&quot;sans-serif&quot;;color:black">Overall level is not= too high (i.e.,=3D =A0no &#8220;roll-over&#8221; effect for intelligibility has occurr= ed).<o:p><=3D /o:p> lt:auto; margin-left:0cm;text-indent:-18.0pt;mso-list:l0 level1 lfo1"> <![if !supportLists]> ans-serif&quot;;color:black"> Ignore">12.&nbsp; <![endif]> t;,&quot;sans-serif&quot;;color:black">We&#8217;re not = including lip readin=3D g.<o:p></o:p> <div> &nbsp; =3D3D"font-family:&quot;Calibri&quot;,&quot;sans-serif&amp= ;quot;;color:black"><o:=3D p></o:p> </div> <div> ;sans-serif&quot;;color:black">These assumptions allow one to, = first, deter=3D mine the AI needed to achieve the assumed target %C given the assumed type = =3D of speech, and, second, determine the SNR required with =A0the assumed spectral shapes to obtain that AI.<o:p></o:p>&lt= ;/span> </div> <div> ;sans-serif&quot;;color:black">&nbsp;<o:p></o:p&gt= ; </div> <div> ;sans-serif&quot;;color:black">For instance, assuming talkers w= ant 95%C wit=3D h &#8220;unfamiliar sentences&#8221; then using the polynomial fits= from Sh=3D erbecoe and Studebaker (JASA 1990) of the ANSI S3.5-1969 transfer functions= =A0between AI and %C, our talkers would need an AI of 0.45. If we assume th= e =3D noise and speech have the same long-term spectral shape then the SNR can be= =3D =A0determined from (SNR&#43;12)/30=3D3D0.45 (staying with the 1969 AI m= ethod), =3D yielding SNR =3D3D 1.5 dB.&nbsp; Subtracting the =A0binaural SNR improvement yields 0.5 dB.&nbsp; <o:p></o:p&gt= ; </div> <div> &nbsp; =3D3D"font-family:&quot;Calibri&quot;,&quot;sans-serif&amp= ;quot;;color:black"><o:=3D p></o:p> </div> <div> <div> ;Tahoma&quot;,&quot;sans-serif&quot;; color:black">From: uot;Tahoma&quot;,&quot;sans-serif&quot;; color:black"> AUDITORY - Research in Auditory Perception [<a hre= f=3D3D"mailto=3D :<a href=3D"mailto:AUDITORY@xxxxxxxx">AUDITORY@xxxxxxxx</a>&q= uot; target=3D3D"_blank">mailto:<a href=3D"mailto:AUDITORY@xxxxxxxx= TS.MCGILL.C">AUDITORY@xxxxxxxx</a>=3D A</a>] On Behalf Of Andy Sabin<o:p></o:p></span&= gt; </div> ;Tahoma&quot;,&quot;sans-serif&quot;; color:black">Sent: uot;Tahoma&quot;,&quot;sans-serif&quot;; color:black"> Wednesday, January 22, 2014 9:53 AM<o:p></o:= p> <div> homa&quot;,&quot;sans-serif&quot;; color:black"> To: <a href=3D3D"mailto:<a href=3D"mailto:AUDITO= RY@xxxxxxxx">AUDITORY@xxxxxxxx</a>" target=3D3D"_bl= ank">AU=3D <a href=3D"mailto:DITORY@xxxxxxxx">DITORY@xxxxxxxx</a></a&= gt; Subject: Reference for typical SNRs is public spaces<= o:p></o:p></spa=3D n> </div> </div> <div> <div> &nbsp; =3D3D"font-family:&quot;Calibri&quot;,&quot;sans-serif&amp= ;quot;;color:black"><o:=3D p></o:p> </div> <div> Hi List,&nbsp;<sp=3D an style=3D3D"font-family:&quot;Calibri&quot;,&quot;sans-s= erif&quot;;color:bl=3D ack"><o:p></o:p> </div> <div> &nbsp; =3D3D"font-family:&quot;Calibri&quot;,&quot;sans-serif&amp= ;quot;;color:black"><o:=3D p></o:p> </div> <div> Can anyone point me to a=3D =A0reference showing SNRs that are typically observed in public spaces (e.g= .,=3D =A0restaurants, bars ...etc)? I can find this info for overall SPL, but am = ha=3D ving a hard time finding it for SNR.&nbsp; :&quot;Calibri&quot;,&quot;sans-serif&quot;; color:black"><o:p></o:p> </div> <div> &nbsp; =3D3D"font-family:&quot;Calibri&quot;,&quot;sans-serif&amp= ;quot;;color:black"><o:=3D p></o:p> </div> <div> Thanks =3D3D"font-family:&quot;Calibri&quot;,&quot;sans-serif&amp= ;quot;;color:black"><o:=3D p></o:p> </div> <div> Andy Sabin tyle=3D3D"font-family:&quot;Calibri&quot;,&quot;sans-serif= &quot;;color:black"=3D ><o:p></o:p> </div> <div> &nbsp; =3D3D"font-family:&quot;Calibri&quot;,&quot;sans-serif&amp= ;quot;;color:black"><o:=3D p></o:p> </div> </div> </div> </div> libri&quot;,&quot;sans-serif&quot;; color:black"><o:p>&nbsp;</o:p></p&g= t; </div> </div> </body> </html> --_000_5801ADC15F4A8D4CAC8F4B3195FA6327EF3B322BSPMXMBX32vumcnl_-- </blockquote></div> </div> --bcaec548a62757d44904f1009bf3--

This message came from the mail archive
/var/www/postings/2014/
maintained by:

DAn Ellis <dpwe@ee.columbia.edu>
Electrical Engineering Dept., Columbia University