On 06/26/2011 10:13 AM, [email protected] wrote:
On Sat, 25 Jun 2011 16:23:29 +0200 (CEST), [email protected]
wrote:
Hi,
Can anyone point me to a simple code example for how to determine the
phase at a specific time in a waveform?
ex. if I have a sample that is 5 seconds long and want to know the
phase
at 2.5 seconds
I'm open to code in any language or a scripted example if such a tool
exists. If there is an ui which has that feature I am also
interested.
in general, theory i mean, phase equals delay; it can be computed
deterministically iif the signal is a pure tone (sinusoidal) otherwise
you must have a reference signal and referring to its fundamental
frequency (1st order harmonic) or, when in-the-field, you can pick one
conventional signal tone (i've been told that a 1khz is a usual choice:)
and compare to that.
if you are to compare two known signals (often input vs. output of some
lti transformation, eg. filter), than the computation to do is
correlation. well, you should use fft for that, but i'm sure you know
that:)
You wouldn't happen to have an example of this code?
Perhaps there is something in qtractor along these lines already?
no, not at all :) qtractor doesn't have anything of the sort, sorry. i
was just clearing dust from the top of my mind re. lti systems theory :)
I have looked at integrating something with librubberband but first things
first right?
hmmm... there's this wsola algorithm for time-stretching in the
time-domain (cf. frequency-domain as in rubberband) where
auto-correlation is computed to find the optimum overlap window point
(by waveform similarity). well, that's maybe an idea spark... but again
you need the zero-phase/origin reference signal waveform anyway to
correlate to... 'coz, as others said, your question doesn't make much
sense as is :)
cheers
--
rncbc aka Rui Nuno Capela
[email protected]
_______________________________________________
Linux-audio-dev mailing list
[email protected]
http://lists.linuxaudio.org/listinfo/linux-audio-dev
