Hi. I have 8 looping arrays, they're all sized 65535 (so 64kb each). They have fades at the start and at the end.

My question is this.

I would like to (have control of the) offset of playback.

So, let's say I have a plucked melodic sound in two arrays. The pluck-sound starts right at the beginning of the array.

I would like to de-sync them (they would currently play at exactly the same speed, so both plucks "pluck" at the same time).

I would like to offset the other. How would I go about doing that? This would let me "kind of" do tape looping where they keep triggering but so that they are slowly going off sync.

@esaruoho said:

@whale-av said:

The sample to be played is now shorter, but you want it played at the same speed (values coming in at the same rate), and to stop when [tabplay~] receives the number [65536]....... and start again.

hmm.. i'm not looking for the sample to be shorter, i'm looking for the sample to be the same length (65536) yet start at, say, "at halftime" of the sample, so if i have two sounds going

Sorry, I should have been more precise "the part of the sample to be played is now shorter".
David.

@bocanegra said:

Leave out the multiplier on the offset. The ramp goes from 0 to 1. The wrap module wraps around 1.

hi, i realizd that there was no reason for me to do any multiplication on the knob input (p8-offset) because it already sends 0...1 so for that to be multiplied by 65536 was useless.

now all i need to do is smoothen the output so it doesn't click. that's next:

thanks for all your assistance thus far! i'll see what whale suggested next now that i can verify that there is an offset occurring in this

@whale-av said:

@esaruoho Use it to duck the signal output of a particular sampler while adjustment numbers are coming in.
Something like.

hmm.. did i get this right? that.. if the sampler receives offsetnumbers, just close the volume out of that sampler and then when it stops, return it back? so in this case, first set [*~]. to 0 for 5 ms (?) then delay 5 ms, then set it back up to 1 in 5 ms?

@esaruoho That is the idea. You could try less than 5msec. It needs to ramp in a time that is not perceived as a click. As you are (almost certainly) jumping on onset to a value that is not zero you need to ramp to it.

Also you might have missed in my previous post that if you are using long samples you should use the double-precision onset inlet of [tabread4~] rather than shifting the phasor, and you could of course use it for all your samples.
David.

@whale-av

i'm using 64kb samples for everything, all arrays are 64kb of size..

i'm now getting a "silence" which then stops and the sound fades back in, when moving a knob. sometimes it crackles, sometimes it doesn't. i wonder how i could make this better.

@esaruoho It gets more complicated then. But not too much.
You will need to use [samphold~].
There is an example F09.declickit.pd in the doc folder.
David.

@esaruoho PS....... you are not using [line~]..... you are using [line].
With [line~] it should be fine.
You will need to remove the "1" from the [*~ 1] of course.
It is very hard for us to test, communicate or make corrections unless you upload your patch...... and extremely hard to spot any errors.
Try this....... Microsound-granular.pd ........... if you get clicks then there could be a different problem..........
David.

@whale-av said:

@esaruoho PS....... you are not using [line~]..... you are using [line].
With [line~] it should be fine.

thing is, where would [line~] be connected to? directly to [+~] or to [tabread4~]?

@bocanegra said:

Sooo, best way is to add the offset as a fraction (offset in samples / total samples) then run the ramp through a [wrap~] before you scale it up with the multiplier. Like this:

[phasor~][/~] <- offset/length
|/
[+~]

Hi @bocanegra - just wondering, what kind of a number would i be putting into [/~] ? if the array is 65536, then should i divide (and it appears to be an audio-divide) the incoming 0...1 numberspace (coming from my midicontroller knob)?

this is the bit that i'm most confused about.

@whale-av said:

It is very hard for us to test, communicate or make corrections unless you upload your patch...... and extremely hard to spot any errors.

hi!
plom.pd

here is the current state. if i have a slider going from 0 to 1 (replacing [r p$1-offset], and twist a knob (on midicontroller) or insert a slider there, i will hear loud sounds. you can load a sample by clicking on the (L) and then set the numberbox underneath it to be 1 (or 2 or whatever). then just replace anything after [*~ 1] as [dac~ 1 2]
anything to the right can be fully ignored.

@esaruoho Just change [line] to [line~] in your patch (post 18 of this thread I think), and set all the timings to 5ms, and remove the 1 from the [*~ 1] which is just a volume control at the end of the sampler that will duck the audio as the click occurs.
The [1 118( message is far too long, and that is why you hear the sound drop out.
If you upload plom.pd here I can correct it and post it again for you.
David.

@whale-av said:

@esaruoho Just change [line] to [line~] in your patch (post 18 of this thread I think), and set all the timings to 5ms, and remove the 1 from the [*~ 1] which is just a volume control at the end of the sampler that will duck the audio as the click occurs.
The [1 118( message is far too long, and that is why you hear the sound drop out.
David.

ok, i moved to below 10ms but this is what i hear
resample 0008 [2018-10-14 184954].mp3

and patch looks like

@esaruoho The output from [r p$1-offset] is going to the [+~] after [phasor]........ before [line~] takes the volume down. You need to delay that data with [pipe 4] so that the volume is at zero as the change is made.
David.

@whale-av said:

@esaruoho The output from [r p$1-offset] is going to the [+~] after [phasor] before line takes the volume down. You need to delay that data with [pipe 4] so that the volume is at zero as the change is made.

hi! i added the [pipe 4] and before that i was messing around with modifying the message data amounts and delay amounts until i got something that felt like there was no dropouts:

so this was with a regular signal going into 0 1( and a delay 10 signal going into 1 30(

i added pipe 4 and it seemed to help. if i take 0 1( and increase it to 0 5( i'll start hearing audible clicks. hmm. this is really getting there, so thanks for your help!

looks like delay 15 to 1 5( and regular signal going into 0 1( and the actual offset-data going through pipe 4 helps. I'll see how this sounds like. i'm hearing drops, which is fine, but no crackles, which is great.

i'm not sure how to rejig this so that the drop would be as minimal as possible, but there would be no stuttery clicks. what's the recipe, please?

@esaruoho Yes, if you go to 0 in 5 ((... [0 5(.... )) with a [pipe 4] then the offset message is sent after 4 msec..... and the [line~] needs another 1msec to get to zero........ because you sent it the message... go to 0 in 5msecs.
David.

@whale-av said:

remove the 1 from the [*~ 1] which is just a volume control at the end of the sampler that will duck the audio as the click occurs.

thing is, if i don't set it to 1, the sampler will not play until a knob has been turned.. so i'm trying to figure out how to force-set that to 1 without using *~ 1 which i normally use for setting volume to 1 by default. i even tried loadbang->bng->1( -> *~ input 2 but that didn't seem to work either

@esaruoho I didn't think of that.....!......
Put the loadbang to a message [1( sent into the left inlet of the [line~]

Or maybe you can leave the 1 in your version of Pd. In Pd Vanilla the [*~ 1] will not let you connect an audio signal to the right inlet, only a control signal.
David.

@whale-av said:

@esaruoho I didn't think of that.....!......
Put the loadbang to a message [1( sent into the left inlet of the [line~]

Or maybe you can leave the 1 in your version of Pd. In Pd Vanilla the [*~ 1] will not let you connect an audio signal to the right inlet, only a control signal.
David.

if i have [*~ 1] on purrdata, i cannot connect a line~ there. i can, however, connect a line there.maybe i should try regular non~ line again?

@esaruoho Sorry, long phone call.

That's what I thought (about [*~ 1]. You could try [line 1 1] which will start with a value of 1, and when you send it a message it will "jump" every ms...... but I think you will still get clicks as the level change will be a "staircase" with 5 steps.
[line~] and the loadbang will be better.
David.