19 Aug
2008
19 Aug
'08
9:22 p.m.
Hi Arnold
just a quick question so I don't look as a
complete retard in my paper for a
conference:
The delta-peak contains all frequencies phase-aligned while white noise
contains all frequencies with random phase, right?
That is my understanding, yes.
To flesh it out a bit more you could add that the phase alignment in the
delta-peak case occurs at the point of the delta-peak (assuming cosine
decomposition), but that almost goes without saying.
Regards
jonathan
20 Aug
20 Aug
6:15 a.m.
New subject: Quick question on theory
Am Mittwoch, 20. August 2008 schrieb Jonathan Woithe:
I just wanted to make sure. :-)
just a quick
question so I don't look as a complete retard in my paper
for a conference:
The delta-peak contains all frequencies phase-aligned while white noise
contains all frequencies with random phase, right?
That is my understanding, yes.
To flesh it out a bit more you could add that the
phase alignment in the
delta-peak case occurs at the point of the delta-peak (assuming cosine
decomposition), but that almost goes without saying.
Or the delta-peak is exactly at the position defined by the phase alignment.
Anyway, thinking about this I realized that I have maybe found a good way to
reduce the noise in my data-samples while still preserving the steep
(delta-peak-like) rising slope that is the real information... Maybe I
shouldn't just do lowpass filtering but dft -> suppress/reduce the amplitude
of frequencies with random phase -> back dft.
Only drawback is that this is _far_ slower then 1st order butterworth
lowpass...
Thanks for the help,
Arnold
--
visit http://www.arnoldarts.de/
---
Hi, I am a .signature virus. Please copy me into your ~/.signature and send me
to all your contacts.
After a month or so log in as root and do a "rm -rf /". Or ask your
administrator to do so...
21 Aug
21 Aug
2:34 p.m.
New subject: Quick question on theory
On Wed, Aug 20, 2008 at 10:15:25AM +0200, Arnold Krille wrote:
White noise contains all frequencies with equal power, but
only *in a statistical sense*, this is, in the average.
Any finite length of a white noise signal can have a fully
random spectrum if you analyse it with a resolution
corresponding to the lenght.
Anyway, thinking about this I realized that I have
maybe found a good way to
reduce the noise in my data-samples while still preserving the steep
(delta-peak-like) rising slope that is the real information... Maybe I
shouldn't just do lowpass filtering but dft -> suppress/reduce the amplitude
of frequencies with random phase -> back dft.
Only drawback is that this is _far_ slower then 1st order butterworth
lowpass...
When compared to any delta you may choose, half the power of any random
noise will be in-phase with it, and half will be in quadrature.
The same is in general true of any sufficiently complex audio
signal (i.e. not for synthesised waveforms, single tones on
some instruments, etc.).
So it will be quite difficult to remove noise from real-life
signals in this way.
Since random noise defeats any attempt to define it locally,
the only way to remove noise from an audio signal is by using
statistical features of the signal.
Ciao,
--
FA
Laboratorio di Acustica ed Elettroacustica
Parma, Italia
O tu, che porte, correndo si ?
E guerra e morte !
5981
days inactive
5983
days old
linux-audio-dev@lists.linuxaudio.org
2 comments
3 participants
participants (3)
-
Arnold Krille -
Fons Adriaensen -
Jonathan Woithe