naturerecordists
[Top] [All Lists]

Re: Microphone Noise Floor Problem

Subject: Re: Microphone Noise Floor Problem
From: "picnet2" picnet2
Date: Sat Sep 12, 2009 2:35 pm ((PDT))
Hi,
     Ive tried a few methods, are you talking about "noise floor" from the =
mic or preamp? or general noise pollution?

For Pollution,

Use baudline for a 65K FFT flat EQ against a known noise source, you can ch=
oose a section in the recording and EQ against that (if your not running li=
nux, virtualbox + ubuntu will work in windows or mac os x to get you going)=
.

Despite such a tool, very careful EQ such as the example Rob provided I bel=
ieve to be the best method, removing noise to preserve just the signal ster=
ilizes the acoustic portrait you took during the recording. If capture of t=
he bird vocalization is paramount maybe switching to a narrow beam shotgun =
mic, or a parabolic dish? and not the NT1A's?

Just a thought, I face the same problem myself - I need a magic acoustic er=
aser for traffic noise.

Then again the obvious solution is take a trip where there is less noise po=
llution in the first place - akin to moving your telescope to a dark(er) si=
te.

BIAS sound soap etc is one tool, never really had much joy with it myself. =
The other is removing the noise via e.g FFT preserving the signal, and addi=
ng your own background noise recorded earlier (providing thats noise free..=
.) - which has been discussed here before I think...

If its purely preamp noise, different recorder is needed that has a EIN low=
er than the noise in your mic (calculated from sensitivity + self noise and=
 max spl see the rane site for a description of that..) - or if its mic noi=
se (from the NT1A????) - then a higher sensitivity is needed, or to the ext=
reme a directional mic array with many elements.

You can also increase sensitivity in specific bands by adding mechanical "E=
Q" around the mic or specially constructed resonator at the specific freque=
ncy of interest. Ive recently designed a system that boosts 4 Khz and 8 Khz=
 by 10 and 8 dB respectively, when the "bump" is removed via EQ at those fr=
equencies the result is very quiet. (The self noise of the mic remains the =
same, just the signal in those regions is boosted by the mechanical coupler=
) - Some times we do the same, by placing a hand behind our ear to listen t=
o quiet sounds...

Another system Ive recently investigated is based on a paper from Neumann i=
ts a log spiral boundary array, this is a log spiral with an electret in th=
e center - it will give around +6dB 'bump' from 1 to 10+Khz, must be quite =
effective at reducing the noise due to the increase in signal.

Kind Regards,
Mike.

--- In  "jmccubbinmd" <mccubbin.astronomy@=
...> wrote:
>
> --- In  Rob Danielson <type@> wrote:
> > Hi JM--
> > Fun analogy to think about. Are there some astrophotographers who
> > insist they are documenting "space" opposed to only identifiable
> > objects? "Noise" is a very rich construction when I think of it as an
> > overlap between perceptual and technical limitations.
> >
> > Here's a QuickTime movie experiment I did with your recording and
> > noise reduction examples:
> > http://tinyurl.com/okwpc6
> >
> > If you have trouble viewing it, I can make an mp3.
> >
> > Rob D.
>
> Noise in an astrophoto image translates into grain.  The great thing abou=
t digial imaging in astrophotography is that we use very crude cameras in c=
omparison to say, my Canon 5D MkII.  I basically capture a raw frame with n=
o processing and no "raw import" utility.  It has hot pixels all over it.  =
Some of my exposure times are measured in hours.  But that's off topic.
>
> I hastily processed that audio, and I'm sure there are artifacts in it.  =
What I would like to know is what are the typical processing tricks people =
use to get the noise out of the audio.  This is going to become more and mo=
re essential as I try to get the really quiet things around.  In the earlie=
r audio of that 18 minute sequence, I can hear mosquito's buzzing around th=
e NT1-A's.  So I'm going to have to learn a few new tricks.
>
> I'll work on that audio with Logic, which tends to have cleaner tools in =
my opinion.  Then I'll see what I can do with it.
>








<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