hi there,

looking at using the looper using a live input with delwrite~ and vd~ with buffer.

I want to adjust the length of the loop on the fly as it is buffering the live input. Keep getting clicks though whenever I adjust the loop length.

is there any way to smooth out the loop length changes so I don't get clicks?
patch attached with a line object added, but that hasn't worked.

any help greatly appreciated

thanks

drew

http://www.pdpatchrepo.info/hurleur/3-4-2-5-loop-generator2.pd

I've played with adding line~ or vline~ instead of sig~ when feeding loop time into vd~buff1 - as patch attached shows.

This stops the clicking, but pitch changes when the loop time adjusted.

should I be looking into using another object to smooth clicks? Where abouts in the in the signal path are the clicks coming from?

thanks

I didn't actually run the patch, but the clicks are probably from amplitude discontinuities. Say a signal goes into the buffer at a certain time. then at however much time the delay is set for that signal comes out of vd~. if you decrease the delay, vd~ will jump to that point in the delay line. These sample values are very unlikely to be the same value, so the output of vd~ jumps. This jump causes a click

edit: try using line~ instead, line only updates at block boundaries (hence the jumping). you'll get some aliasing but it's hard to prevent

personally I would use tables/arrays for this, they're more versatile

thanks for the info,

will look into writing and reading simultaneously from arrays/tables continuously - any pointers on where to start reading? I've done a basic search but haven't come up with anything so far - Want to keep it as a live looper continuously writing (recording) and reading (playing) with time constantly and freely changing...

much appreciated

drew

if you have cyclone you can check out this example using poke~ http://puredata.hurleur.com/sujet-5021-sound-sound-looper-clear-option but you still may get clicks with it.
I didn't read your second post carefully enough; when you change delay time there will be phase modulation (doppler effect) because you are upsampling the signal, so the pitch will increase or decrease by however fast you change the delay time. as a side note, see the help patch G09 in the audio examples (doc folder) for how this is used to make a pitch shifter

so what do you need/ what should it sound like?
if you don't need to change the delay time continuously and the buffer is in an array/table you could grab the current amplitude value of the audio signal, the value you will jump to and use miller's switch & ramp in order to fix the discontinuity. you would still get slight clicking but it would be a discontinuity in the derivative rather than the signal so not as bad
might take quite a bit of work though...

you could still use delay lines if there were some way to read out of a delay line in the message domain. a final note: you could see if you can send the change in delay time to a second delread~ first in order to use that value in switch and ramp.
also when implementing the ramp use the value that was in the array before the current audio sample was written

another option is fading the audio out as the delay time is changed and fading it in afterwards

I have revisited this problem, and am still struggling with it.
I did find a max/msp patch that loops live signal without clicks, and I have attached an image of that (excuse my ignorance if thats not the done thing).

I tried to model my patch, which is also attached within the compressed file - but the result is not the same - I don't really understand the [patcherargs] object in max, even after referencing help etc... and not sure what it does for this patch, or if there is something similar in pd.

It feels like it's on the right track but just needs a better understanding. Also, the sustain/feedback seems different in the pd version. Not so effective for some reason.

If anyone has any thoughts or guidance, much appreciated

drew

http://www.pdpatchrepo.info/hurleur/live_looper_comparison.zip

Hi

Have a look at the G09.pitchshift in the 3.audio.examples section of the help browser.

hvm

this should be more equivalent, haven't tested if it works though

http://www.pdpatchrepo.info/hurleur/loop-generator2.pd

@seb-harmonik.ar is there a way to do this using arrays?

the patch I posted specifically? You can make your own delay line with arrays but if you're using vanilla the actual buffer (array) size has to be a multiple of 64 because that's the rate that [tabwrite~] gets bangs
and the amount of delay time you can actually use is the array size - audio block size because if it's greater then the writer and the reader will overlap each other.

@seb-harmonik.ar

the patch I posted specifically?

yeah

had been trying to adapt it

so far i have resized the arrays
but still having trouble reading back the samples correctly to avoid clicks and getting the feedback section to work.

@youaresound loop-generator2-array.pd should work @ 44100 & 64 block size

@seb-harmonik.ar hey thanks so much for this.
that is what i was trying to do.