Raimund:
Thank you for confirming what I expected. The key is that the noise picked
up by each of the two mikes is "uncorrelated", hence the phase cancellation=
s
and a 3 db drop when they are mixed together. Same applies, I suppose, to
the noise floor of the mikes themselves, which would be audible on very
still morning if the gain is pushed high.
This is comforting news for those wanting to do high quality stereo
parabolic recordings.
Lang
Lang,
I think that your measurement results are indeed very precise and nicely
confirm a basic law in signal processing, that says that independent noise
signals add geometrically (sum =3D sqrt(a*a+b*b). If the two signals have t=
he
same
amplitude (e.g. 1 V), then the sum of two signals is 1.41 V (the square roo=
t
of 2.0 V) and not 2.0 V as one would perhaps expect. In other words, the
resulting noise level drops by the factor 0.7 ( =3D 1.41 / 2.0), which exac=
tly
corresponds to -3 dB. The fact that you really got that 3dB difference is
also a
proof for the good acoustic isolation between the two mikes. The two mikes
pick up noise from different sources that are not correlated to each other.
The same principle applies to the inherent noise sources within the recorde=
r
(the two preamplifiers and AD converters). Therefore, splitting a monoaural
signal to a stereo input will also decrease the self-noise level (when the
two channels are finally mixed together). However, I guess, that this
benefit
can be almost neglected because the self-noise level of a state-of-the-art
recorder is usually small compared to that of the microphone.
Raimund
> Re: Stereo Parabola versus Monaural Parabola
>
> This morning I did a fairly thorough comparison of the output of my stere=
o
> parabolic setup and the output of my monaural parabolic setup.
>
> My stereo setup was my Telinga Parabola outfitted with two Senheisser MKH
> 20
> omni mikes placed to each side of a plastic barrier just like the design
> that Klas uses in his DAT Stereo mike. I provided pics of my design some
> months back. For those new to our group, the pics may be found here:
>
> http://www.naturesound.com/telinga/telinga.html
>
> My monuaral setup was a Telinga Parabola with a Telinga Universal Mike
> Mount
> outfitted with a single Senheisser MKH 20 omni mike.
>
> My question was whether or not a stereo recording obtained from my stereo
> parabola has a signal-to-noise ratio that is comparable to that of a
> monaural recording from my monaural parabola.
>
> So here's what I did:
>
> To produce a standard signal for my test, I chose my best recording of a
> winter wren and put a single song on a CD. I used noise reduction to
> reduce
> the background of this recording to almost nothing. I went to a quiet
> woodland setting near Ithaca, placed my portable CD recorder on the groun=
d
> and fed the output to a small speaker, using the repeat function of the
> player so that the song would play over and over again, with a pause of
> about five seconds in between. There was a very gentle breeze, insuring
> that
> the ambience heard behind my test recordings would be the slight whooshy
> or
> hissy sound of the gentle breeze, not the mike noise floor, and not the
> ambience from the CD recording itself.
>
> Using the two different setups (stereo parabola and monaural parabola), I
> recorded the winter wren song and adjusted my recorder in both instances
> so
> that the peak record volume was -3b.
>
> After making recordings using both setups, I returned to my studio and
> inputted the test recordings into my computer.
>
> Using PEAK audio editing software, I first chose a song recorded with the
> stereo setup plus some of the gentle breeze ambience that accompanied it.
> I
> then created a monaural soundfile by mixing the two tracks together. Then
> I
> normalized this recording so that the song peaked exactly at -3db. Then I
> deleted the song itself, leaving about five seconds of the gentle breeze
> ambience.
>
> I then processed a monaural recording using exactly the same procedure. B=
e
> aware that the monaural recording was made to two tracks using a Y-cable
> that splits the output of the monaural Telinga into two inputs for the DA=
T
> recorder. So first I mixed the two tracks to create a monaural track in
> PEAK. Then I normalized to -3b. Next I removed the song itself, leaving
> about five seconds of ambience.
>
> Then I made a new soundfile in PEAK, which included both examples of the
> ambience, one following the other with a fraction of a second of silence
> in
> between.
>
> Finally, I compared the loudness of these two samples of ambience with th=
e
> assumption that this would give me an accurate measure of the
> signal-to-noise capability of the two parabola setups when they are used
> to
> create monaural recordings ("noise" here refers to the level of the
> ambience, not mike noise).
>
> The results surprised me. The ambience from the processed stereo recordin=
g
> was exactly 3 db quieter than the ambience from the monaural recording!! =
I
> tried several examples and got exactly the same result each time.
>
> This is good news because it means that recording in stereo using Klas's
> design yields a two-track recording that can be used to produce a monaura=
l
> recording that is every bit as good as what I can get using a standard
> monaural parabolic setup. In fact, the result was 3 db better in terms of
> signal-to-noise.
>
> I had actually expected the opposite: that recording in stereo would
> produce
> a nice-sounding result, but at the expense of some signal-to-noise in
> comparison to a monaural recording (as long as both are made using the
> same
> parabolic reflector, the same microphone type, and Klas's stereo and
> monaural types of design).
>
> What do you tech-heads think of this result? Does this make any sense?
> Don't
> you agree this is good news?
>
> One question I have is whether or not splitting the monaural parabola
> output
> to two channels for DAT recording is influencing the results. I know that
> this could result in a slight increase in the noise floor created by the
> mike, but it shouldn't affect the level of the ambience which accompanies
> the bird song. Using my monaural setup, I typically record to two tracks
> because it sounds better when monitoring (my recorder only allows for
> stereo
> monitoring), and because it provides two tracks just in case there's a
> glitch in one. Anyway, when the two tracks are mixed back to one track in
> the studio, the results should be the same as when only one track is
> recorded from the parabola.
>
> Any thoughts about my test and my conclusions?
>
> Lang
>
>
> "Microphones are not ears,
> Loudspeakers are not birds,
> A listening room is not nature."
> Klas Strandberg
> Yahoo! Groups Links
>
>
>
>=20
>
[Non-text portions of this message have been removed]
________________________________________________________________________
________________________________________________________________________
|