posted 22 August 2010 13:55            

any tips?

posted 22 August 2010 14:35            

i haven't explored the tau as much as i should have.. and i'll look at SoS....

posted 22 August 2010 17:21            

Have you heard my kwerk synthesis example. It used SOS stretched to the extreme (10 sec to 15 mins) which just sounded like boringly very slow changing sine waves, but I then added the kwerk of my voice going AHH (looped) to give it an organic sound again.

The kwerk is the spectrum of the analyzed AHH minus the smoothed AMP and smoothed FREQ version of the same analysis so that the AHH was totally removed and just the deviations from the norm were left.

This on it's own fed to a SOS makes no sound but the deviations can then be added to any other spectrum so that it has the pitch and timbre of that new spectrum but keeps the live feel of a complex waveform provided by the non stretched looped AHH deviations.

Hope this makes sense

Pete

posted 22 August 2010 20:20            

of course, i'd expect kyma to be able to do something like this.. but maybe it can't?

posted 23 August 2010 02:06            

paulstrech is a FFT -> iFFT algo. with transposition etc. which sounds very static, but this should be very easy to create in kyma too if you need it. and of course the fun with kyma is about "realtime"...

if you want to create some "ambient" like sounds like on the famous 800% timestreched justin bieber track... i would go for the crossfilter and maybe a "jessica simpson track" as an response file ;-)

all the best
chris

posted 23 August 2010 11:47            

quote:

---

Originally posted by ChristianSchloesser:

paulstrech is a FFT -> iFFT algo. with transposition etc. which sounds very static, but this should be very easy to create in kyma too if you need it. and of course the fun with kyma is about "realtime"...

---

Oh, really? I assumed it was some kind of granular synthesis, I might have to give this a go in Kyma. Having Kyma do these types of effects in real time is very exciting. Instant ambient!

posted 23 August 2010 19:01            

Your joke about crossing Justin Bieber with Jessica Simpson gave me an idea for a weird way to do time-stretching with the CrossFilter.

Here's an example that is about 5000X time-stretched.

So the idea is to use a looping sample as the Response. Then, set the CrossFilter to capture a short snippet (0.2 s) once every loop time. Except that the loop time is the duration of the entire sample PLUS some small amount of time (0.02 s) so the place it is grabbing the snippet from is gradually moving forward in time through the original sample.

In the first example, I used white noise as the input, so it is a little bit similar in concept to the SampleCloud (and very uniform & smooth, not grainy). But the cool thing about doing it this way is that you can use *different* inputs to excite that little snippet of the sample and get some very interesting results!

(Thanks to Pete Johnston, the CrossFilter can continuously grab new pieces of the Response instead of just being a boring, static response! )

posted 23 August 2010 22:02            

super clever of you, as always.

chris, yeah, i guess the timeindex parameter would simply be the key to a time stretch... i should have realized that...!

posted 25 August 2010 21:07            

quote:

---

Originally posted by pete:
Have you heard my kwerk synthesis example. It used SOS stretched to the extreme (10 sec to 15 mins) which just sounded like boringly very slow changing sine waves, but I then added the kwerk of my voice going AHH (looped) to give it an organic sound again.

---