Is there some kind of 1/4 samp shift per second going on?

Question

I'm currently working on some smoothed random curves and encountered some very strange behaviour. Here is the Sound I'm talking about: http://kyma.symbolicsound.com/qa/?qa=blob&qa_blobid=9516736016941304319

The thing is when I start the Sound the first 4 seconds everything is fine but then it seems like the Delay (1 samp) isn't delaying the input anymore... You can tell because of the spike in the curve. See this screenshot:

 

 

 

 

 

 

 

 

 

 

 

The left channel is the smoothed noise and the right channel is the interpolation function (I'm doing cosine interpolation here but you can switch the Sound to linear interpolation and get the very same strange behaviour).

Now if I increase the Delay (1 samp) to 2 samp of delay time I get the same thing after 8 seconds:

 

 

 

 

 

 

 

 

 

 

 

So it seems like there is a 1/4 samp shift going on here? How can I get around this? Or am I doing something wrong here?

EDIT: I just changed the whole thing to work at 0.5 s and at first I thought it works now... But after 2 minutes I get the spikes again...

Thanks!

Answer 1

By using TriggeredSampleAndHold, (rather than SampleAndHold as you have here), you can create the interpolated random numbers at any rate and adjust it while it's playing.  Since all the sampling is triggered off the master OscillatorTimeIndex, there is no possibility for jitter.

Comments

Very clever! Great! Thanks!

---

Neat, thanks SSC!

Answer 2

What time index are you using for you interpolation?

Is it from a wavetable or capytalk? Is it sample rate?

Comments

I'm using the OscillatorTimeIndex to create the interpolation function. I guess that one is from a wavetable but the Sound doesn't tell. It ought to be samplerate. You can download the Sound by clicking the link above and have a look if you like! Thanks!