Patrick Keenan

I have been making a patch for a composition in which various instruments will be recording into various buffers over the course of 30 seconds. The texture is very sparse so there will be a lot of silence in these buffers.

Later in the composition i want to granulate sections of the audio in these buffers, however i want it all to happen automaticaly without me having to go through the buffers and pinpoint the areas where there is sound for the granulation patch to do its work. However if i just told the granulation patch to process random sections of the buffer i would get a lot of granulated silence which is not what i want.

I came up with what i thought was a solution. I have a [count~] going into an [index~] reading through the buffer and i have the left output of [index~] going into a [number~] object which is then going into an [abs] object and finally into a [>]. By setting the threshold of the[>] i can get it to output a 1 as soon as the sample value in the buffer goes above a certain volume. I then have the patch stop the count object at the exact moment it finds a note onset and send that exact point and however many milliseconds added onto that point to the granuation patch setting the low and high settings for where in the buffer gets processed.

This seems to work but when i get about four or five of these modules going it seems to stop finding the note onsets.

Here is the patch. Sorry it is such a mess but the part of importance is in the bottom right corner, can anyone tel me a better way of doing what i am trying to do?

I have been making a patch for a composition in which various instruments will be recording into various buffers over the course of 30 seconds.  The texture is very sparse so there will be a lot of silence in these buffers.  

Later in the composition i want to granulate sections of the audio in these buffers, however i want it all to happen automaticaly without me having to go through the buffers and pinpoint the areas where there is sound for the granulation patch to do its work.  However if i just told the granulation patch to process random sections of the buffer i would get a lot of granulated silence which is not what i want.

I came up with what i thought was a solution.  I have a [count~] going into an [index~] reading through the buffer and i have the left output of [index~] going into a [number~] object which is then going into an [abs] object and finally into a [>].  By setting the threshold of the[>] i can get it to output a 1 as soon as the sample value in the buffer goes above a certain volume.  I then have the patch stop the count object at the exact moment it finds a note onset and send that exact point and however many milliseconds added onto that point to the granuation patch setting the low and high settings for where in the buffer gets processed.

Here is the patch.  Sorry it is such a mess but the part of importance is in the bottom right corner, can anyone tel me a better way of doing what i am trying to do?

index-reading-through-buffer-finding-onset-of-notes-problems

If it were me, I'd either have [uzi] cycle through [peek~] once the audio is recorded, writing only those samples that are above your threshold into a new [buffer~] which can then be used by as many playback objects as you like, OR.. have this threshold detection done in realtime; ie, have another [buffer~] set up to record only those fragments that exceed a certain value while the instruments are being recorded.

Index reading through buffer finding onset of notes problems!