commons-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Al Chou <hotfusion...@yahoo.com>
Subject Re: [math] transform - fft documentation bug? proposal window function & sonogram
Date Tue, 04 Dec 2007 08:34:30 GMT
Not having ever looked at our FFT code, I won't address the issue of whether the documentation
is right or wrong (I suspect the documentation simply has a typo; base-2 FFT's are the easiest
to implement, given that they're the only kind anyone ever seems to write about, so I can't
believe we have a base-4 implementation, which would be less easily usable).

I'm +1 on providing window functions, given how commonly needed they are.  I'm -0 on the spectrogram,
having personally spent time trying to use a naive spectrogram as a power spectrum estimator.
 I only spent enough time on that task to come to the understanding that it wasn't as simple
as it seemed it should be.  Maybe even that understanding is wrong.  In any case, I would
prefer that the package provide a function that "does what most people think of when thinking
about a spectrogram" rather than the naive spectrogram.  Unless my understanding, as I mentioned,
is wrong and the naive spectrogram is actually useful.


Al
-- 
Albert Davidson Chou 
    Get answers to Mac questions at http://www.Mac-Mgrs.org/ .

----- Original Message ----
From: Carsten Kr├╝ger <C.Krueger@gmx.org>
To: dev@commons.apache.org
Sent: Monday, December 3, 2007 5:45:03 AM
Subject: [math] transform - fft documentation bug? proposal window function & sonogram


Hello,

the class documentation[1] of fft states
"We require the length of data set to be power of 2"
and the method documenation[1] this
"Perform the base-4 Cooley-Tukey FFT algorithm (including inverse)."

One must be wrong:
base-4 alogrithm means that the length of data has to be power of 4

fft without window function is pretty useless, it would be nice to
implement typical[2] window functions for example:
-Hamming
-Hann
-Blackman
-Trinangle
-Rectangle
-Welch
-Gaussian

after having fft and windowfunction, sonogram/spectrogram[3] would be
nice and is easy to implement (10 lines of matlab code).
for example:
public double[][] sonogram (int fftsize, int overlap_factor, int
 windowfunction)

greetings
Carsten

[1]
http://commons.apache.org/math/apidocs/org/apache/commons/math/transform/FastFourierTransformer.html

[2]
http://en.wikipedia.org/wiki/Window_function

[3]
http://en.wikipedia.org/wiki/Spectrogram


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@commons.apache.org
For additional commands, e-mail: dev-help@commons.apache.org






      ____________________________________________________________________________________
Get easy, one-click access to your favorites. 
Make Yahoo! your homepage.
http://www.yahoo.com/r/hs 
Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message