Subject: Re: Phoneme categorization in noise From: Matt Winn <mwinn83@xxxxxxxx> Date: Wed, 15 Feb 2012 08:15:53 -0500 List-Archive:<http://lists.mcgill.ca/scripts/wa.exe?LIST=AUDITORY>--f46d0435c20add57c804b90082f5 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: quoted-printable Daniela, Another way to increase difficulty is to use a vocoder. There are multiple options for vocoding, but the basic idea is to replace portions of the spectrum with carriers comprised of sinewaves or noise bands, resulting in the loss of spectral resolution. Those carriers can either be matched (in center frequency) to the spectral regions that they represent, or they can be shifted in frequency (as is the case in some cochlear implant simulations), or they can be compressed in frequency range. All of these changes result in speech that is difficult (but not impossible) to perceive. A common "classic" citation for noise band vocoding is Shannon et al. (1995). Another more recent one with more details about the number of spectral and temporal bands is by Xu et al. (2005). Another one I like that incorporates frequency compression is Ba=C5=9Fkent and Shannon (2003). Ther= e are many more papers that address these techniques, but these are straightforward and easy to understand. Here are the references: Ba=C5=9Fkent, D., & Shannon, R. (2003). Speech recognition under conditions= of frequency-place compression and expansion. *Journal of the Acoustical Society of America, 113*, 2064=E2=80=932076. Shannon, R., Zeng, F-G., Kamath, V., Wygonski, J., & Ekelid, M. (1995). Speech recognition with primarily temporal cues.=E2=80=9D *Science,* *270*= , 303=E2=80=93304. Xu, L., Thompson, K. & Pfingst, B. (2005). Relative contributions of spectral and temporal cues for phoneme recognition. *Journal of the Acoustical Society of America, 117,* 3255-3267. On Wed, Feb 15, 2012 at 7:00 AM, Daniela Sammler <sammler@xxxxxxxx> wrote= : > Dear list, > > I need to increase the difficulty of a phoneme categorization task (e.g. > "ba" vs "pa"). One possibility would be to present the phonemes in noise = to > receive a shallower psychometric function. > > Is this the way to go or is there any other/better way to increase > difficulty in phoneme categorization? > > Thank you very much! > Kind regards, > Daniela > > > -- > Daniela SAMMLER, Ph.D. > Max Planck Institute for > Human Cognitive and Brain Sciences > Stephanstr. 1a > 04103 Leipzig > phone: +49 341 9940 2679 > fax: +49 341 9940 2260 > --f46d0435c20add57c804b90082f5 Content-Type: text/html; charset=UTF-8 Content-Transfer-Encoding: quoted-printable Daniela,<div>Another way to increase difficulty is to use a vocoder. There = are multiple options for vocoding, but the basic idea is to replace portion= s of the spectrum with carriers comprised of sinewaves or noise bands, resu= lting in the loss of spectral resolution. Those carriers can either be matc= hed (in center frequency) to the spectral regions that they represent, or t= hey can be shifted in frequency (as is the case in some cochlear implant si= mulations), or they can be compressed in frequency range. All of these chan= ges result in speech that is difficult (but not impossible) to perceive.=C2= =A0</div> <div>A common "classic" citation for noise band vocoding is Shann= on et al. (1995). Another more recent one with more details about the numbe= r of spectral and temporal bands is by Xu et al. (2005). Another one I like= that incorporates frequency compression is=C2=A0Ba=C5=9Fkent and Shannon (= 2003). There are many more papers that address these techniques, but these = are straightforward and easy to understand. Here are the references:</div> <div><br></div><div><p class=3D"MsoNormal" style=3D"margin-left:27pt">Ba=C5= =9Fkent, D., & Shannon, R. (<span>2003)</span>. Speech recognition unde= r conditions of frequency-place compression and expansion.=C2=A0<i>Journal = of the Acoustical Society of America, 113</i>, 2064=E2=80=932076.</p> <p class=3D"MsoNormal" style=3D"margin-left:27pt"><br></p><p class=3D"MsoNo= rmal" style=3D"margin-left:27pt">Shannon, R., Zeng, F-G., Kamath, V., Wygon= ski, J., & Ekelid, M. (1995). Speech recognition with primarily tempora= l cues.=E2=80=9D<span>=C2=A0=C2=A0</span><i>Science,</i>=C2=A0<i>270</i>, 3= 03=E2=80=93304.</p> <p class=3D"MsoNormal" style=3D"margin-left:27pt"><br></p><p class=3D"MsoNo= rmal" style=3D"margin-left:27pt">Xu, L., Thompson, K. & Pfingst, B. (20= 05). Relative contributions of spectral and temporal cues for phoneme recog= nition.=C2=A0<i>Journal of the Acoustical Society of America, 117,</i>=C2= =A03255-3267.</p> </div><div><br></div><br><div class=3D"gmail_quote">On Wed, Feb 15, 2012 at= 7:00 AM, Daniela Sammler <span dir=3D"ltr"><<a href=3D"mailto:sammler@xxxxxxxx= bs.mpg.de">sammler@xxxxxxxx</a>></span> wrote:<br><blockquote class=3D= "gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;padding= -left:1ex"> Dear list,<br> <br> I need to increase the difficulty of a phoneme categorization task (e.g. &q= uot;ba" vs "pa"). One possibility would be to present the ph= onemes in noise to receive a shallower psychometric function.<br> <br> Is this the way to go or is there any other/better way to increase difficul= ty in phoneme categorization?<br> <br> Thank you very much!<br> Kind regards,<br> Daniela<br> <span class=3D"HOEnZb"><font color=3D"#888888"><br> <br> --<br> Daniela SAMMLER, Ph.D.<br> Max Planck Institute for<br> Human Cognitive and Brain Sciences<br> Stephanstr. 1a<br> 04103 Leipzig<br> phone: <a href=3D"tel:%2B49%20341%209940%202679" value=3D"+4934199402679">+= 49 341 9940 2679</a><br> fax: =C2=A0 <a href=3D"tel:%2B49%20341%209940%202260" value=3D"+49341994022= 60">+49 341 9940 2260</a><br> </font></span></blockquote></div><br> --f46d0435c20add57c804b90082f5--