Hi there,

I'm using a 3D accelerometer to alter soundtracks, modulating sound effects, or recognize some events to launch a a given sound.

Human motion data (besides impacts) is mostly within 0-12 Hz. Now I would like to lowpass the accelerometer data (about 2 to 4 Hz) to use the values as a kind of inclinometer getting rid of the segment accelerations and keeping the orientation.

The accelermeter data are sent via wifi I´ve set a data rate (banging the read-in) at 5 Hz. Maybe the low acc data rate is the problem.

I´ve tried using [lop] but there is not much information on the tutorial. My idea was to put the acc signal into the hot plug and write the cut off frequency inside the box as an argument, but this does not seem to work. By the way, the non audio [lop] has 4 inlets, which only makes my confusion bigger.

Using [mavg] helps to some extend, but what I´m trying is to get rid of "rapid changes in the data" i.e. the high frequency content.

Any ideas out there?

Any help is appreciated.

Cheers,

Gaspar

I use the [mapping/smooth] with my iPod acc.

Thanks a lot Gromit,

I´ve just played a bit around with it and it seems promising for my purposes.

Now I will need some more fine tuning work and hopefully I can get what I want.

Cheers,

Gaspar

PD: The nice thing about the phones, is that you can read out directly the orientation data and don´t need to go through the acc. values.

I also found [mapping/iir] quite useful for this purpose.

Thank you James,

I´ll have a look at it as well. In the mean time I found [median_n] quite interesting. In the past I used median filters a lot (for different stuff). The nice thing about them besides their simplicity is that outliers or short peaks do not affect the output.

In any case, for the moment I´ve solved my problem without filtering using conditionals [expr if...] at precise moments. Since this seems to work for this particular thing I´ll go with that. But sooner or later I´ll have to come back to this.

G.