From: Markus M. <me...@me...> - 2005-07-06 06:49:12
|
Hi Martyn, this is awesome!! Do you have a patch file against CVS HEAD for that, so I can integrate it? Markus Mar...@ao... schrieb: > Hi all, > > Particularly those interested in the problem with the Equalizer that > was reported back in April and discussed at length (particularly > window lengths!). > > I have located the root cause of the problem and why it shows up in > 'Equalizer' but not in 'FFT Filter'. I believe (and can see) that my > fix also fixes the 'Spectrum' view which (I had not reported this as > an error and don't know if anybody else had) gives spurious horizontal > lines even when you generate a perfect sinusoid. > > The core of the FFT routine uses 'float' when it should be using > 'double' (at least). The build up of errors for longer block lengths > is significant and causes the problems. 'FFT Filter' uses a (hard > coded) length of 256, and the problems are insignificant whereas > 'Equalizer' uses a (hard coded) block length of 16384, which causes > significant build up of errors, it would appear. I have not done > analysis to find out which of the variables is most significant and > boldly changed about 11 of them. > > I have given more details at http://members.aol.com/MartynShaw/ and a > fixed windows exe. I don't mind discussing my findings should you have > any queries at all. > > HTH > > Martyn > > PS Are there any other 'signal processing' type bugs/errors/queries > that I can get my teeth into? |