posted 21 June 2006 05:43            

I guess that you want to record into ram in a linear (normal possibly looped) way and then read it off in a random fashion. You would use the step writer if it was important that the signal is recorded randomly not just played back randomly. you may be better off using the up down counter to feed the index of a waveshaper reading from ram. you could fill the ram with a normal kyma write to disc module. Let me know if I'm understanding correctly?

posted 25 June 2006 11:33            

I was just looking for alternatives to solve this problem and hoped that the StepWriter might be an option (?)
But maybe I'm better off by cutting all those words in advance in an audio editor, singling them out (incl. the rests between them), then
loading them into a multi-sample and playing them with a random index (utilizing a PulseTrain as its gate/trigger, taking the max. word size as its min. period length or something like this)?
The next possible answer creates the next questions :-)
Thanks!
Cheers, flo

posted 26 June 2006 03:17            

Do I understand the effect correctly. You start with the recording of a poem. You cut the poem into clips on each word border. You swap the word clips around but time stretch or compress (so as to maintain the same pitch ) to get the duration to match the original words duration . You do some funky tricks to make sure that the joins between words are not so clicky by either cross fading or pitch smoothing and formant smoothing. You end up replacing the words with the original words so that the original poem is heard.

Is this it?

posted 29 June 2006 04:05            

no, ‘same duration’ just applies to an unchanging entire duration. Maybe I can better explain this with one sentence first (i think the basic idea is much simpler and that time-, pitch, formant-shifting is not necessary):

1. subdivide a sentence (speech recording) into its words, incl. the rest behind it – the rest between this word and the next word (one could single/cut out the rests too, but let’s keep it simple)

2. then gradually go from a randomized order sentence to the semantically correct order, and that on a sentence-basis (throw the dices only once for every sentence); that is, one sentence with all its words (and rests), but gradually per sentence less and less puzzled (randomly ordered), until the original semantically correct sentence is arrived at;
Also important: within those sentences no word (sample) is allowed to occur more than once, as within the original, thus only the order of words (incl. rests) should be randomized, and therefore the entire duration of those sentences is always the same.

Everything puzzled now?

thanks again.
Best, flo

posted 30 June 2006 20:11            

quote:

---

Originally posted by SSC:
If you are doing the word ID and cutting ahead of time, you could turn them into separate samples, put them all into the same folder and use MultiSample with a random Index.

---

A couple of years ago I attempted almost the exact same effect with the MultiSample. If all the samples are the same duration, it works fine.

The problem I had was that my samples were not the same length, and there was no way to know when a sample finished playing so I could then randomly select the next one.