audio units - question about conv and splitting load
audio units - question about conv and splitting load
- Subject: audio units - question about conv and splitting load
- From: Jeff Lieberman <email@hidden>
- Date: Sat, 28 May 2005 18:21:09 -0400
hi-
i'm just getting started with an audio unit exploring some
research i've recently done, and haven't made one before,
so i'm hitting some problems. currently i'm grabbing beat/tempo
information and buffering an entire [8 bar] looping sample...
so first i check when the beat resets, and start buffering so i
have a generally aligned buffer. i then take ffts of the little
128 sample chunks that i have, to generate a spectrogram.
here's where i'm at. i need to convolve this spectrogram with a
filter. but for me to convolve the whole thing at once is impossible,
i think it will halt the system too much.
1) is there a way to split up the load for an operation so it will
automatically do a certain amount of processing on the whole
thing, and then wait? right now what i want to do is convolve each
little section, and save all the parts and overlap them later.
2) given that thought, i'm having a problem with conv - right now
it's always returning an array of length inputlength-filterlength.
but really i want what i think should be inputlength+filterlength-1...
[ie i think i should be also getting a tail from where the parts only
partially overlap].
then i'd be able to overlap all the parts. i tried to find more
documentation
on conv - but it just mentions things like 'filterStride' without
mention
as to what those really do..
any help would be much appreciated.
jeff
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Coreaudio-api mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden