Hey guys!

I'm trying to use bonk~ to detect different kinds of touches on a touch screen (tap, knock, punch, etc..), but I'm having some trouble with analyzing the signal that comes from the microphone (Im doing this on an Android tablet).

I started playing around with thresh and debounce objects to see if I could tune bonk~ to only listen to sound every 200ms or so (because no one will tap faster on the touch screen) to eliminate some of the noise and false positive results.

Can you help me better understand threshold object and if I can use it to do what I explained?
This microphone is really sensitive but also not very good, so I can't mess around much with the signal (filter ambient noise and voices and such is not very easy for me also )

Any input is appreciated!

Here is my patch: testRec.pd

Hi,

I started playing around with thresh and debounce objects to see if I could tune bonk~ to only listen to sound every 200ms or so (because no one will tap faster on the touch screen) to eliminate some of the noise and false positive results.

To do this, probably it would be easier to work with [onebang] after the [bonk~] object:

Cheers!
Gilberto

Hey! Thanks for answering so fast

I've read about what you're saying, and tried it, but still dont get what I want.

In the end I want to be able to get only (or for the most part) the sound that a touch on a touch screen produces, to distinguish them. Any ideias or pointers to do this?
With a contact microphone it would be easier, but I want to rely only on the tablet built-in microphone..

Thanks in advance!

Hmm, I see (and that's indeed a bit trickier).

What about using some sort of threshold control between the [adc~] and the [bonk~]? Something like this:

filtering-signal.pd

I had created this patch for as an attack detection for a composition of mine, and you may be able to adapt it for your needs.

Cheers,
Gilberto

Hi

You can train bonk~ to react on different hits. See the help file.

mvh

@gsagostinho Hm that's becoming too complex for me! ahaha but thanks.

I'm trying to use low and high pass filters to reduce some of the noise. However, still not getting nice results (even training bonk~ a lot I can't seem to get precise results in detecting different touches everytime)..

@PedroPT sorry to dig this, but for guitar chords (that need a lot of debounce) I found using a [thresh 15 17( message was doing the trick