"Jim Gort" <jgort@comcast.net> wrote in message news:<qvO7c.65736$SR1.114965@attbi_s04>...> Rune: > > I didn't completely understand it while in DSP at MIT, but I got the gist > that Cepstrum analysis is the core of homomorphic deconvolution, which has > many real-world benefits such as recovering original data from scratched > LPs, etc. Thus, though I have never used it in my work as an EE, I think it > *may* be an important part of signal processing that ought not lightly be > dismissed as "a curiostity". Others could comment much better than I--my > career is a hardware designer and DSP coder, not a DSP theorist > (unfortunately). > > JimYou are right, of course. The point I tried to make, was that cepstrum can be very powerful when it can be used. However, that power comes at a price in terms of the cepstrum apparently being very sensitive to prior conditions. If the problem does not match perfectly with the required conditions, the cepstrum does not work. I did some searches in IEEExplore when I researched these things. If you search for "cepstrum <or> homomorphic" you find some 700 articles. And that's not a lot for such a generic search in a 30+ year time span. So while people know of the cepstrum, not every many appear to use it. I have been told the cepstrum is important in speech processing, and can be used for certain types of image enhancement, but that's about it. So in summary, I dientified certain problems with the cepstrum myself, and I see certain indications in the literature database that there may be problems with this technique. Rune
Minimum phase filter design using cepstrum methods?
Started by ●March 17, 2004
Reply by ●March 23, 20042004-03-23
Reply by ●March 23, 20042004-03-23
"robert bristow-johnson" in news:BC808AC0.99B3%rbj@surfglobal.net...> > *everyone* doing DSP and wanting to deal with and understand algorithms > should have a copy of *some* O&S, in my opinion. >Hi Robert. Al Oppenheim used to go further than you do. To those of us who worked around him he advocated buying a copy for the car, as gifts, etc. etc.
Reply by ●March 23, 20042004-03-23
In article 105vqblo66iphe6@corp.supernews.com, Max Hauser at maxREMOVE@THIStdl.com wrote on 03/23/2004 02:38:> "robert bristow-johnson" in news:BC808AC0.99B3%rbj@surfglobal.net... >> >> *everyone* doing DSP and wanting to deal with and understand algorithms >> should have a copy of *some* O&S, in my opinion. >> > > Hi Robert. Al Oppenheim used to go further than you do. To those of us who > worked around him he advocated buying a copy for the car, as gifts, etc. > etc.sounds like he was scraping for royalties. r b-j
Reply by ●March 25, 20042004-03-25
Hello Michael, I did handle ( similar one ) this problem in my P.hD thesis. Issue is how does one knows, "for given data what is necessary and sufficient condition that provides satble and minimum phase transfer function". In fact, we could get only sufficient condition and not necessary condition. May be good to focus on this hard problem and may be they can look for P.hD degree for thier contirbution. Thanks in advance. Kind regards jk http://www.epigon.co.in/ "Michael Numminen" <michael.numminen@comhem.se> wrote in message news:<ZTs6c.86740$dP1.248790@newsc.telia.net>...> Hi Ron! > > To your question, see below. > > "Ronald H. Nicholson Jr." <rhn@mauve.rahul.net> skrev i meddelandet > news:c3ahbu$5cp$1@blue.rahul.net... > > I've run across a few web pages describing how to convert an arbitrary > > FIR filter to a minimum phase variant by the use of cepstral methods: > > e.g. > > > wn = [ones(1,m); 2*ones((n+odd)/2-1,m) ; ones(1-rem(n,2),m); > > > zeros((n+od d)/2-1,m)]; > > > y = real(ifft(exp(fft(wn.*real(ifft(log(abs(fft(x))))))))); > > etc. > > > > I've tried this algorithm on a few windowed-sinc filters and it seems > > to work... much more straight-forward than trying to find the zeros > > of a z-transform and move them inside. > > > > Why does this technique work? How does a cepstrum help generate a > > minimum phase filter? > > > I don't remember exactly becous it was so long time ago now. But the reason > why it works depends on that the cepstrum is a logarithmic function. And, if > I remember correctly the cepstrum has a properity of minimum phase. You have > to search in the theory > of Analytical Functions or Theory of Complex Functions, here you can find > correct answear. > > Best regards, > > Michael Numminen, M.Sc.E.E. > > > Thanks. > > -- > > Ron Nicholson rhn AT nicholson DOT com http://www.nicholson.com/rhn/ > > #include <canonical.disclaimer> // only my own opinions, etc.
