Subject: [linux-audio-user] notch/peak? filters ladspa question.
From: Marco Scoffier (marco4linux_AT_earthlink.net)
Date: Tue Jun 04 2002 - 08:59:19 EEST
Recently I have been cleaning up some audio files and have wanted to
remove a particular frequency. In the past, I have used tools where I
have a 0db line in the middle of the graph : top +18db: bottom -18db.
I can pull control points to either create a peak or a notch and vary
the width with a variable Q 1(v.wide)->100(v.narrow). This was
particularly handy in removing a constant hum b/c I would make a very
narrow peak up to +18db scan over the frequencies until I heard a
distinct whistle then move the curve to -18db and the annoying
frequency could be removed or made less annoying.
There is a similar window in snd but I am only able to remove
frequencies, I can't do the original test. And the bottom line doesn't
have a scale which relates to Hz. Once you have found an offensive
frequency if it is a hum, you can often remove more by going to the
harmonics, doubling and halving, so having the frequencies across the
bottom is handy.
Using ecasound I made a notch by splitting my sound into two channels
and highpassing one and lowpassing the other then remixing the output.
-a:1 -efh:4000 -a:2 -efl:4500 -a:1,2 -o I could make a peak with the
Frankly I get quite confused with the filters. The resonance doesn't
really correspond to the Q. I guess I'm asking for an explaination a
bit like the jimmies if you know ircam's library, to make biquad~ and
co. a bit more perceptual. Has there been any work towards this? Or
explainations about how the "DSP" parameters like number of samples per
feedlooop, or resonance should be used. What do the ranges correspond
The ladspa plugins suffer from this sort of confusion about what the
ranges are actually doing. Perhaps I haven't been reading the right
thanks in advance for your input,
-- "for the people" only works in conjunction with "by the people." the MPAA is buying congress. do something today.
This archive was generated by hypermail 2b28 : Tue Jun 04 2002 - 08:45:13 EEST