how to eliminate jitter in timesliced channels

I have two timesliced channels which represent an xy position, which (since it comes from a trace sop) have some jitter in them that I’d like to eliminate
losing as little as possible of the larger movements.

I couldn’t figure out using the current tools what/how to use. I’m sure there’s good filters, any suggestion is welcome.

thanks!
dani

Have you tried the “filter” chop?

the problem with the filter chop is that it changes the data, acting pretty much like a lag (if you want it to have some effect).

what I need is something that eliminates low level noise smoothly. I.e. where the small up/down changes really cancel each other out and make something really smooth (this is basically noise, so it varies from frame to frame, should be easy to eliminate) but as the changes (say in 2-3 frames) begin to outline a certain direction, the tendency of the channel should be to start to get affected by them. Kind of like a spline function where the bigger the change the more keys it gets, but small changes can be lost and smoothed out.

d

I’m not a signal processing expert, but I don’t see how you would do this without some lag. Like how do you interpret between a 0 0 0 0 0 1 2 0 from a 0 0 0 0 1 2 4 before the last sample? Is the 1 2 signal or noise?

I would think that - I’m going to say this in intuitive, not academic terms - signal variations could be interpreted as forces applied to mass. If the object/signal is at rest, alternating positive and negative push (noisy signal) wouldn’t really affect the output that much, but as signals take strength the output begins to move and it’s friction lessens as its momentum increases. it’s not exactly the same thing, but the idea is different than a simple filter, which doesn’t take into account the current motion vector. it’s basically a filter that’s lessened exponentially as the signal moves, and it’s increased dramatically when it’s not.

Yes there’s a little lag, but it’s mostly to gain momentum. I guess it’s like a particle system approach to signal processing. mmm… maybe I can do that with particles. I only need a couple anyway.