From 45ac408d0abed030eaf8f84af031c0de86e5f5b0 Mon Sep 17 00:00:00 2001 From: Sebastian Dröge Date: Sun, 19 Aug 2007 19:16:33 +0000 Subject: gst/filter/: Add small comparision with the chebyshev filters in the docs. Original commit message from CVS: * gst/filter/gstbpwsinc.c: * gst/filter/gstlpwsinc.c: Add small comparision with the chebyshev filters in the docs. --- gst/audiofx/audiowsincband.c | 5 +++++ gst/audiofx/audiowsinclimit.c | 7 +++++-- 2 files changed, 10 insertions(+), 2 deletions(-) (limited to 'gst/audiofx') diff --git a/gst/audiofx/audiowsincband.c b/gst/audiofx/audiowsincband.c index 79fa39d9..bf399cbc 100644 --- a/gst/audiofx/audiowsincband.c +++ b/gst/audiofx/audiowsincband.c @@ -46,6 +46,11 @@ * controls rolloff and stopband attenuation. The Hamming window provides a faster rolloff but a bit * worse stopband attenuation, the other way around for the Blackman window. * + * + * This element has the advantage over the Chebyshev bandpass and bandreject filter that it has + * a much better rolloff when using a larger kernel size and almost linear phase. The only + * disadvantage is the much slower execution time with larger kernels. + * * Example launch line * * diff --git a/gst/audiofx/audiowsinclimit.c b/gst/audiofx/audiowsinclimit.c index 3cf14e5d..0db1a743 100644 --- a/gst/audiofx/audiowsinclimit.c +++ b/gst/audiofx/audiowsinclimit.c @@ -31,8 +31,6 @@ * is probably the bottleneck * - Maybe allow cascading the filter to get a better stopband attenuation. * Can be done by convolving a filter kernel with itself - * - Drop the first kernel_length/2 samples and append the same number of - * samples on EOS as the first few samples are essentialy zero. */ /** @@ -46,6 +44,11 @@ * controls rolloff and stopband attenuation. The Hamming window provides a faster rolloff but a bit * worse stopband attenuation, the other way around for the Blackman window. * + * + * This element has the advantage over the Chebyshev lowpass and highpass filter that it has + * a much better rolloff when using a larger kernel size and almost linear phase. The only + * disadvantage is the much slower execution time with larger kernels. + * * Example launch line * * -- cgit