Subject: [AUDITORY] Lyon hearing book: free PDF now updated and announced to the public From: "Richard F. Lyon" <dicklyon@xxxxxxxx> Date: Tue, 2 Jan 2018 14:33:52 -0800 List-Archive:<http://lists.mcgill.ca/scripts/wa.exe?LIST=AUDITORY>--f403045c2220842e940561d2b1a5 Content-Type: text/plain; charset="UTF-8" I've made a 2018 version of my book with all the 2017 errata corrected, and announced the free PDF on my blog site http://www.machinehearing.org/ And I deleted the robots.txt file that asked crawlers not to index it, so it should show up in web searches eventually, too. Actually, there's one more error that I've been discussing with Volker Hohmann, but can't easily fix with tiny edits: some of the things I said about discrete-time gammatones and impulse-invariance and pole-zero-matching methods are not quite right. Probably I'll try to explain this in a blog post, in case anywhere cares for such subtleties. In particular, it means that Angela Darling's discrete-time gammatone implementation does not have an impulse response that is exactly a sampled version of the continuous-time gammatone, at least for orders greater than 2. Hohmann settled on Darling's method (before knowing of it) as an excellent approximation, while I had mistakenly claimed it was exact. Darling had said "A digital IIR filter is designed using the impulse invariance method to implement the Gamma Tone filter...", which was true, sort of, but her approach was to apply the impulse invariance method to the cascaded stage filter, not to the overall filter, and I had failed to appreciate the subtle difference, which showed up when Hohmann applied impulse invariance correctly to the complete gammatone. Dick --f403045c2220842e940561d2b1a5 Content-Type: text/html; charset="UTF-8" Content-Transfer-Encoding: quoted-printable <div dir=3D"ltr"><div><div>I've made a 2018 version of my book with all= the 2017 errata corrected, and announced the free PDF on my blog site <a h= ref=3D"http://www.machinehearing.org/">http://www.machinehearing.org/</a><b= r><br></div>And I deleted the robots.txt file that asked crawlers not to in= dex it, so it should show up in web searches eventually, too.</div><div><br= ></div><div>Actually, there's one more error that I've been discuss= ing with Volker Hohmann, but can't easily fix with tiny edits:=C2=A0 so= me of the things I said about discrete-time gammatones and impulse-invarian= ce and pole-zero-matching methods are not quite right.=C2=A0 Probably I'= ;ll try to explain this in a blog post, in case anywhere cares for such sub= tleties.=C2=A0 In particular, it means that Angela Darling's discrete-t= ime gammatone implementation does not have an impulse response that is exac= tly a sampled version of the continuous-time gammatone, at least for orders= greater than 2.=C2=A0 Hohmann settled on Darling's method (before know= ing of it) as an excellent approximation, while I had mistakenly claimed it= was exact.=C2=A0 Darling had said "A digital IIR filter is designed u= sing the impulse invariance method to implement the Gamma Tone filter...&qu= ot;, which was true, sort of, but her approach was to apply the impulse inv= ariance method to the cascaded stage filter, not to the overall filter, and= I had failed to appreciate the subtle difference, which showed up when Hoh= mann applied impulse invariance correctly to the complete gammatone.<br></d= iv><div><br></div>Dick<br><br></div> --f403045c2220842e940561d2b1a5--