i want to accompany a small timelapse with audio recorded in much the same way as momentary photos in a timelapse.

i want to record a half second of audio every 15 seconds (should be a variable) into a buffer and then when finnished ouput a wave file with all the bits put together to form a continouos line of sound.
i have no idea how it would sound.

anybody have any ideas?

You could try and record it directly to disk using [writesf~]. Stick it inside of a subpatch with a [switch~], and use a [metro 15000] to turn the [switch~] on every 15 seconds and turn it off a half second later using a [delay 500] or [makenote 1 500]:

[metro 15000]
|
[makenote 1 500]
|
[switch~]

The [switch~] turns dsp on and off in the subpatch, making it act sort of like a pause/play. You'll probably want to use a declicking envelope as well.

I had similar idea, and made a patch for timelapse recording, see the attached file.
As Maelstorm suggested, declicking envelope is essential for making a smooth recording, but I couldn't solve that problem. I need to find a way to ramp the audio up and down just after the start and prior the stop of recording. You'll see that I tried to make it by placing the [line~] object after the [adc~], but it still gives me audible clicks.
Also, pay attention to the sample rate you are working with, If you are on 44100 Hz, just change it in [expr] object, and adjust the table length (44100 * number of seconds).

http://www.pdpatchrepo.info/hurleur/timelapse_rec.pd

i didn't have a look at the patches, but you should get that kind of declicking envelope by simply using vline~ i think, since every period is the same length of 500ms. feed it with [0, 1 5, 0 5 495( and you have an envelope with a linear attack time of 5 ms and a decay time of 5 ms 495ms after the bang. bang this message every time switch~ receives a 1 and multiply vline~s output with the signal before it goes into writesf~.

[line~] is probably better than [vline~] here because with both solutions the recording starts and stops at block boundaries. [line~] will line up to those boundaries, but [vline~] won't.

Edit: well, actually, you could use a [bang~] + [spigot~] combo to trigger the message to [vline~] and line it up to the block boundary.

In fact, [delay] doesn't line up to the block boundaries, either. I'm not certain, but this might be why you're still getting clicks in your patch, sinewave.

@ maelstorm: jap, you're right, i didn't think about that.
@ sinewave: you should substract the decay time of 20 ms from the delaytime for the first [delay] (in the example it should be 480 ms not 500 ms) - otherwise you always record 520 ms of audio but you start the new chunk after 500 ms. i fixed that just by inserting a [- 20] object befor the [del] object - i don't get any clicks here.

Thank you guys, now the clicks are gone.