spear sdif format ????
i am trying to create a non-realtime video render engine which reacts to sound.
so i set out to analyse music with spear then load the sdif file into sdif-buffer and step thru the frames once a video frame was rendered according to the sound analysis date.
however the data structure that spear produces looks very different to the one of e.g. "moanin.add.sdif" which comes with the cnmat objects (unfortunatelly i don/t know with wihch software it was created)
whereas moanin.add.sdif looks like this:
tuples: 1.000000 99.628510 0.000776
tuples: 2.000000 237.471481 0.046275
...
tuples: 79.000000 9207.314453 0.000640
tuples: 80.000000 9318.450195 0.000804
and the one i did with spear like this:
tuples: 2.000000 6851.086914 0.000131
tuples: 3.000000 6208.314453 0.000463
tuples: 8.000000 1592.039185 0.005841
...
tuples: 680.000000 2244.308594 0.001874
tuples: 681.000000 15669.262695 0.000081
tuples: 684.000000
rest of message:
the frequency column of the spear analysis is not ascending but really mixed up.
while this doesn't seem to be a problem for threefates and sinusoids~ it gives me some trouble in retreiving amplitude values for certain freqencies.
hope somebody can help me!
thnanks, michael
Quote: sm wrote on Sat, 30 June 2007 19:05
----------------------------------------------------
> i am trying to create a non-realtime video render engine which reacts to sound.
> so i set out to analyse music with spear then load the sdif file into sdif-buffer and step thru the frames once a video frame was rendered according to the sound analysis date.
>
> however the data structure that spear produces looks very different to the one of e.g. "moanin.add.sdif" which comes with the cnmat objects (unfortunatelly i don/t know with wihch software it was created)
> whereas moanin.add.sdif looks like this:
> tuples: 1.000000 99.628510 0.000776
> tuples: 2.000000 237.471481 0.046275
> ...
> tuples: 79.000000 9207.314453 0.000640
> tuples: 80.000000 9318.450195 0.000804
>
> and the one i did with spear like this:
> tuples: 2.000000 6851.086914 0.000131
> tuples: 3.000000 6208.314453 0.000463
> tuples: 8.000000 1592.039185 0.005841
> ...
> tuples: 680.000000 2244.308594 0.001874
> tuples: 681.000000 15669.262695 0.000081
> tuples: 684.000000
>
>
----------------------------------------------------
On 30 Jun 2007, at 19:20, sm wrote:
>
> rest of message:
> the frequency column of the spear analysis is not ascending but
> really mixed up.
> while this doesn't seem to be a problem for threefates and
> sinusoids~ it gives me some trouble in retreiving amplitude values
> for certain freqencies.
> hope somebody can help me!
don't think i can help you, but maybe explain a few things.
the first number in every row is the index number of the sinusoidal
track, followed by freq, amp and phase.
as you observed the track numbers (ids) normally start off with low
numbers. but as the sound evolves, new partials are added and others
may disappear. the numbering of the partial tracks is left to the
analysis software - some always use higher numbers for new tracks,
some reuse track ids of partials that have disappeared again - but
the numbers don't have any meaning beyond the information for the
synthesis software, wether to create a new partial or to keep track
of the same partial from the preceeding frame.
does that make any sense to you?
if you want something like ascending order of the frequency column,
you would have to sort it yourself (which surely is possible, but
employs some overhead naturally).
i don't know what software cnmat used for moanin.add.trc, but i might
have read somewhere that it was tuned "by hand" afterwards.
volker.
i could use the files created with spear in cnmat object without big problems.
only when i try to access the data of individual frames, e.g. to see if a certain frequency range is above or below a threshold, i get into troubles since the order of the frequency tracks seems to be arbitrary.
i hoped that there would be trick in spear to get the freqs into a nice ascending order.