naturerecordists
[Top] [All Lists]

Re: Producing clear sonograms

Subject: Re: Producing clear sonograms
From: "" animalsounds
Date: Fri Jun 16, 2006 1:06 am (PDT)
I concur with Gianni and Steve that the level threshold and dynamic range (=
gray level scaling for the  various sound levels) will greatly influence th=
e visual clarity of a spectrogram. Due to the limitations of the analog pri=
nting technology and the electronics of the ancient Kay Sonagraph machines,=
 the dynamic range of those sonagrams was pretty small. I guess it was less=
 than 35 dB (which means that sound components below -35dBFS were hidden). =
Nowadays, digital spectrograms can display dynamic ranges well beyond 100 d=
B. However, I believe that such large dynamic ranges are not useful if you =
just want to display the essential properties of an animal sound. On the ot=
her hand, a large dynamic range could be useful for detecting very soft sou=
nd components or for evaluating the background noise floor.

Another parameter that influences the clarity of a spectrogram is the tempo=
ral resolution. The old Kay Sonagraph machines supported two analysis bandw=
idth settings: narrow (45 Hz) or wide (300 Hz). Bird song researchers usual=
ly used the wide setting. This relatively high bandwidth provides a good te=
mporal resolution (1/300Hz =3D 3.3 ms), which preserves the rapid frequency=
 modulations that are present in many bird songs. The users of digital spec=
trogram software tend to use inappropriately small analysis bandwidth setti=
ngs (they often believe that large FFT size numbers were always an advantag=
e). However, a FFT size of 512 points with a Hamming window and a sample ra=
te of 22.05 kHz provides a bandwidth of 56 Hz only, which will smear the te=
mporal details of the sound (temporal resolution =3D 18 ms). So, one should=
 increase the bandwidth by selecting a smaller FFT size, a better window sh=
ape (FlatTop instead of Hamming) and/or by applying zero padding (reducing =
the frame size).

You might try this out with the free Avisoft-SASLab Light software:
http://www.avisoft.com/downloa_.htm or http://www.avisoft.com/avisoftdemo.e=
xe

Use the command =E2=80=9CAnalyze=E2=80=9D/=E2=80=9DSpectrogram Parameters..=
.=E2=80=9D and try different FFT and Frame sizes. The resulting Bandwidth i=
s displayed at the top of this dialog box.

Regards,
Raimund

--


Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen!
Ideal f=C3=BCr Modem und ISDN: http://www.gmx.net/de/go/smartsurfer




<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