any chances updating the wave array to hold more than 1000?

MichaelBenjamin · 5 May 2010

just after sending this i had the idea to use more than 1 wave array with selectors. might work.

infuzion · 5 May 2010

Making a massive WaveTable bank are we?

MichaelBenjamin · 5 May 2010

yes, mainly for singlecycle and/or quite short waves, since there is also that memory limit. i wish i could just dump some wave folder into it without these restrictions.

mwvdlee · 5 May 2010

+1 for me, though about a "medium" priority. There are more pressing issues I'd like to see resolved (i.e. the envelope retrigger issue and the comma-delimiter thing with text-to-array).

I'm not quite upto 1000 yet (just short of 400 at the moment), but I can certainly see me scaling up to well beyond a 1000 in the future. If I do some quick math, the highest number I can envision right now would be about 2000, so obviously 4000 would still be too little for some 😉

Using multiple wave arrays is a bad choice (though, with no alternatives, it's really the only choice you have). I've tried splitting up my wave array to clean up my schematic but the performance suffered badly.

oddson · 5 May 2010

MichaelBenjamin wrote: yes, mainly for singlecycle and/or quite short waves, since there is also that memory limit. i wish i could just dump some wave folder into it without these restrictions.

Excuse me if I haven't thought this through enough... if it's for single cycles or short wave snippets couldn't you concatenate them together and use start offsets to read them out?

I can imaging it would add some unwanted complication but it shouldn't impact performance significantly.

exonerate · 5 May 2010

oddson wrote: MichaelBenjamin wrote: yes, mainly for singlecycle and/or quite short waves, since there is also that memory limit. i wish i could just dump some wave folder into it without these restrictions.

Excuse me if I haven't thought this through enough... if it's for single cycles or short wave snippets couldn't you concatenate them together and use start offsets to read them out?

I can imaging it would add some unwanted complication but it shouldn't impact performance significantly.

I think for that method creating a schematic which concatenates the single cycles automatically might be a good idea. 😉

Unless there is an app out there that will already do this?

And I doubt it would impact performance at all doing it that way instead of with wave arrays.

Cheers 😉

MichaelBenjamin · 5 May 2010

Excuse me if I haven't thought this through enough... if it's for single cycles or short wave snippets couldn't you concatenate them together and use start offsets to read them out?

I can imaging it would add some unwanted complication but it shouldn't impact performance significantly.

nice when you use words i have to look up. there are mostly the same 500 words on the net. so, about concatenating them - this would imo be too much of a pain, since they are not the same size. i cut them out of music and various sources, plus there are also short one shot samples.. so i would have to make a list with all sample sizes, then another list with the start /end points for each sample inside the merged one and that for about 3000 small/tiny samples. this would be too much for me. sure it could be automated somehow but why not just raise the memory/index restrictions on the array module?

btw the internal storage also seems to waste half the memory since it stores in 32 bit, when only 16bit is needed.

mwvdlee · 5 May 2010

I agree increasing the limit would be the best fix (and 1000 sounds like an arbitrary limit, so probably easy to change), but just out of curiosity... couldn't the concatenation be done efficiently at runtime in some way?

exonerate · 5 May 2010

mwvdlee wrote: I agree increasing the limit would be the best fix (and 1000 sounds like an arbitrary limit, so probably easy to change), but just out of curiosity... couldn't the concatenation be done efficiently at runtime in some way?

Yes can be done.. I've created a module HERE. You will have to test efficiency and see if it is up to scratch. But I think it should prove fast enough.

Cheers 😉

Exo

oddson · 5 May 2010

MichaelBenjamin wrote: nice when you use words i have to look up

I have a tendency to use somewhat obscure words, it's more a predilection towards garrulity than an propensity for obfuscation. 😉 But 'concatenate' is a word I've only ever encountered in computer programming. Because of this (and that it's an Excel function) I assume most of the tech-heads here would know it already.

MichaelBenjamin wrote: this would imo be too much of a pain, since they are not the same size. i cut them out of music and various sources, plus there are also short one shot samples.. so i would have to make a list with all sample sizes, then another list with the start /end points for each sample inside the merged one and that for about 3000 small/tiny samples. this would be too much for me. sure it could be automated somehow but why not just raise the memory/index restrictions on the array module?

A single array could hold start offsets; the entries at i and i+1 combined can give you the start, end and length... but I take your point.

I guess I was assuming from your earlier post you were using a fixed size (1024, 2048) to store single-cycle waveforms. For this I think it could work well.