naturerecordists
[Top] [All Lists]

Re: How would you clean up this file?

Subject: Re: How would you clean up this file?
From: "Dana Blackmer" danablackmerphd
Date: Tue Jun 27, 2006 7:30 pm (PDT)
Thanks Walt, Rob, Derek, Ed, Tim, Curt and anyone else I may have missed
who responded to my questions.  It was a terrific learning experience for
me and gives me a lot to think about and experiment with for a while.  I
hope to be able to do more of this in the future and hope that others were=

able to get something out of this dialogue.  It really helps less
experienced people like me to have people who are so generous with their
time and expertise.  I greatly appreciate it.

Thanks again,

Dana

At 09:36 PM 6/27/2006 -0400, you wrote:

>Posted by: "Dana Blackmer"
> >
> > Thanks, Walt. I have a few questions:
> >
> > I get the 80Hz noise in all my recordings. Is this "normal noise" for a
> > Portadisc and MKH 30/40? Also, is the "low shelf filter" you used to kn=
ock
> > it out equivalent to using an FFT filter in Adobe Audition?
> >
> > I have the same questions about the 18kHz noise.
>
>I think the filters are probably similar. The shelf filters are set for
>a target frequency, gain level plus or minus, and the sharpness of the
>cutoff in dB per octave. The result is cutting off all frequencies below
>or above the target depending on if it's the low or high shelf. The same
>filter could be used to introduce gain for the shelf, of course. They
>are also sometimes called cutoff filters.
>
>The noise at low frequencies does not appear to me to be due to the
>portadisc or MKH mics. It varies from none to very strong and it's
>primary source is distant man made sounds in my experience. Or distant
>storms, waterfalls, whatever. The lower the frequency the farther it
>carries so the higher frequencies of these sources have been filtered
>out. One of the sources at my house is a gas turbine power plant that's
>5 miles away. I can tell that one as it's run for peak electricity loads
>only so can detect when it's running or not. It's quite loud here in the
>low frequencies. Though only somewhat audible in general terms.
>
>With my MKH-110 SASS I can see even greater levels down into the
>infrasound. The MKH-110 has a design low frequency spec of 1Hz, though
>the portadisc limits out before that.
>
> > Concerning the distortion noise that shouldn't be there - can you descr=
ibe
> > it or tell me where it is on the sonogram? I didn't identify it and wou=
ld
> > like to be able to hear/see it as you did. If you worked on the file fr=
om
> > my original post (DawnChorus.mp3) the distortion may have occurred beca=
use
> > I may have inadvertently encoded the MS signal twice, using Audition's
> > Channel Mixer's "Mid-Side to L-R" pre-set. I later posted an original f=
ile
> > (www.danablackmer.com/AudioFiles/DawnChorusMS.wav) and an mp3 file
> > (www.danablackmer.com/AudioFiles/DawnChorusXY.mp3) that I know wasn't
> > double encoded.
>
>It's mixed throughout the sonogram in bands. The sound is sort of like
>water running down a fairly large drain. Also some calls have a sort of
>warble to them.
>
>The wav file is way too big for me to download through the modem, but I
>listened to the mp3 version. The distorted sound is gone. And in it's
>place is more insect sound, so that's probably what was messed up the most=
.
>
>Careful fiddling with the noise filter might allow some toning down of
>the insects, but it's tricky. Their frequencies are mixed through the
>frequencies of the calls you want. Insects can be a big problem here as
>the summer goes on. Really reduce how much reach you have for other sounds=
.
>
>Sometimes where you have a sharp band of insect sound and nothing else
>at that frequency you can use a notch filter or a very narrow parametric
>filter to take it down.
>
>With all filtering listen very carefully. And do sonograms. It's so easy
>to cut into what you want to keep. And any heavy filter application
>frequently results in new sounds produced by the filter. A light touch,
>sometimes applied more than once, generally works best.
>
>When first learning filtering do a awful lot of just playing with
>filters. Try all kinds of settings. When I first had to get serious
>about filtering for the GA frog CD, I spent over a month just playing
>with the files with filters before saving anything. What sounded good
>one day might sound poor the next.
>
>I also prefer to do critical filtering with headphones rather than
>speakers. Though I'm sure there are differences of opinion on that. My
>reasoning is I'm not having to deal with the room ambiance where I'm
>working on top of the site ambiance that I'm trying to preserve. And
>headphones generally color sound less than most speaker systems. Of
>course I do check out the results in speakers too.
>
> > You also said that you added and trained a denoise filter (Spark XL TC
> > Denoise) to decrease this distortion noise. Would this be similar to
> > Audition's Noise Reduction filter (a filter that works by capturing a
> noise
> > reduction profile from a segment of about 0.5-1" of noise-only sound an=
d
> > then using that to profile to reduce the noise in the entire recording)=
.
>
>Should be similar, though each filter differs slightly depending on how
>the programmer wrote it. I used a 500ms sample length, though it's
>settable down to just a few ms. The noise reduction filter I use has a
>audition function where you can listen to the sounds that will be
>removed as you play the piece. Helps if you are trying to preserve some
>sound. I sampled until the green frog calls would be removed minimally
>and there was the least change in the sonogram. The sonogram just looked
>cleaner.
>
>Walt
>
>







<Prev in Thread] Current Thread [Next in Thread>
Admin

The University of NSW School of Computer and Engineering takes no responsibility for the contents of this archive. It is purely a compilation of material sent by many people to the naturerecordists mailing list. It has not been checked for accuracy nor its content verified in any way. If you wish to get material removed from the archive or have other queries about the archive e-mail Andrew Taylor at this address: andrewt@cse.unsw.EDU.AU