specgram6.2/spectrogram7.1
- From: "Salviati" <eckard.blumschein@xxxxxxxx>
- Date: Mon, 4 Feb 2008 06:46:02 +0100
While 'specgram' in MATLAB version 6.2 showed a partially wrong but highly
interesting specrogram for NFFT=64, NOVERLAP=63 and an input similar to that
I used in my IEEE-paper http://home.arcor.de/eckard.blumschein/M283.html,
the corresponding 'spectrogramm' in the corrected version 7.1 excluded
overlap in excess of 50%. As a result of this correction one can for any set
of parameters only get either a narrowband picture with just horizontal
lines, exhibiting frequency even for time to come in a non-causal manner, or
a wideband picture with just vertical lines. While I argue that spectrograms
are dirty mathematics in general, except for near to 100% overlap, the
lacking gradual transition between constant frequency and constant time has
been proven wrong within every mammal's ear. In physics, the stupid
alternative view was dramatized in 1935 by Schroedinger and Einstein after
Heisenberg speculated in 1927: "I believe that the existence of the
classical path can be formulates as follows: The path comes into existence
only after we observe it."
As soon as quantum computers are not just available but they also fulfil
what one expects based on quantum entanglement, I will apologize for
possibly being wrong. I consider it the duty of mathematicians to clarify
what happens due to transition to infinity in case of Fourier transform or,
as I prefer, cosine transform. However, mathematics seems still to suffer
from the obviously wrong and abandoned at last in 1923 naive set theory and
from the Hilbert space footing on it. Naive set theory and Hilbert space are
still considered correct regardless that in 1935 v. Neumann himself lost his
belief in Hilbert space.
I am posting to comp.soft-sys.matlab because I hope that the current or
future version of MATLAB will allow to visualize the gradual transition
between time and frequency or position and momentum, respectively. Who is
responsible for the signal processing part?
Maybe, there is already an other software available that does this job
better. I do not know MAPLE. Aren't there other many commercial or
individual spectrograms too?
Salviati
.
- Prev by Date: Re: urgent quenstions about GADS
- Next by Date: Re: Application Error on Mathlab R15 Startup
- Previous by thread: backpropagation-target setting
- Next by thread: Figure window of video stream to stay active
- Index(es):
Relevant Pages
|
