[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Question about band-passed noise by Fatima Husain

Dear  Colleague, 

I came accross two reactions to your request about how to get
appropriate bandpass noises.  The most direct and clever way is to
generate the signals yourself.  That was what you did and on which my
friend and teammate Yoshitaka Nakajima elaborated elegantly in his
reaction.  Sometimes, my sensei tends to forget that he is very
clever, and therefore I will add four easy alternatives (#2 through
#5).  If, however, # 1 fits you best, I apologize for bothering you
with the less intelligent ones.  But I am sure many colleagues are or
will become confronted with analogous problems, and therefore I
attempted to mail this reaction to the general list instead of to you
privately (if I did not succeed, please put it in public).

There are at least four less elegant solutions to your question:
2) Buy a hardware filter (I used the brand Krohnhite,
Butterworth-type, in my old days), feed in white noise (for instance
from Coulbourn Instruments Modules - do they still exist??? -, or
better from Bruel & Kjaer, Denmark, Europe, which I could not afford)
and turn the dial knobs on the front panels to the preferred bandpass
frequency positions.  Yes, this is really done manually, unbelievable
nowadays: turning knobs.  A very simple but expensive solution,
specially when considering the skirt of the filter (24 dB per octave,
if I recall well).

3) Purchase commercial software: Matlab Professional and one of its
add ons: the Signal Processing Toolbox (The Mathworks Company has
special prices for educational institutes!!  In the Netherlands, I
paid approximately Dfl. 2,000 for the bunch, please convert yourself
into US $, must be thousand or so).  You will be really astonished by
the versatility of signal generating (all kind of waveforms),
filtering (Bessel, Butterworth, Chebyshev, Ellip) and signal analyzing
options (oscillogram, spectrogram, power spectral density plot, and of
a graphical beauty people always dreamed of to put in their
web-blurb).  Almost too much luxury for an experimental psychologist.
I'm very enthusastic about the combination, in particular because you
can simply send your sound vector to any cheap soundcard like
Soundblaster (though I use Turtle Beach cards as well) by the Matlab
playwave statement.  In addition, the latest version of Matlab (v.5)
and Signal Processing Toolbox (v.2), just released, handles 16 bit
sounds, and stereo.  Poor guy: I still go mono at 8 bit.  (Thus,
there's no need to buy any of those ridiculous expensive A/D-D/A
boards with 8 or 16 channels of which we only need two.)  In addition,
and this is probably my best message, you can run on line auditorys
own programming language as well, similar to Basic (credits go to my
young bright students who really did it).

4) Register for inexpensive shareware: GoldWave, developed by Chris
Craig in New Foundland, is the ideal tool.  Latest version 3.22,
features, among many other things, all kinds of signal generating
(there's an expression evaluator, looks like a calculator, by which
you can make sounds you never thought of, and filtering options (low
pass, high pass, bandpass, bandstop(notch)).  The steepness of the
skirts ca and I suspect by viewing the ripple in the passband, that
the developer picked ellip filters, but it's a pity he doesn't mention
the filter type.  Nevertheless, it's a fantastic tool, and it's just
around your corner: Boston is quite close to New Foundland from my
perspective (the Netherlands ?? Is that in Amsterdam??).  To be frank,
I use Goldwave in tandem with Matlab & its Signal Processing Toolbox
(because they communicate extremely well), thus combining inexpensive
and expensive tools (another interpretation of 'Dutch treat' ??).

5) Buy an Audio CD with 1/3 octave band noises, and sample them.
Yoshitaka Nakajima brought the Audio Test CD-1 from the Japan Audio
Society to my place (Leiden University, the Netherlands), and
sometimes I simply sample the track of the CD I need, and of course by
....... GoldWave.

The second reaction to your request I read was by Biao Tian, and is
equivalent to solutions # 3,4 albeit that these are Windows instead of

For your information, I enclose the Websites of Matlab and Goldwave,

http://web.cs.mun.ca/~chris3/         (E-mail: chris3@cs.mun.ca)

Information about the audio CD:
Audio test CD-1 (91 test signals for home and laboratory use) produced by:
14-34 Jingumae
1 Shibuya-ku,  Tokyo 150, Japan

Best regards and success with your research program,

Gert   ten Hoopen
Faculty of Social & Behavioural Sciences
Unit of Experimental & Theoretical Psychology
Leiden University
PO Box 9555,  2300 RB  Leiden,  the Netherlands
Phone: +31 71 527 3871
Facsimile: +31 71 527 3619
E-mail: TENHOOPEN@rulfsw.leidenuni

[manually converted to plain text by dpwe, 1997jun21]