Mike S

I've been experimenting with pitch~ in conjunction with a mic input to find the F0 of the incoming signal. The problem I have is it will give an accurate result then flip to a strange one just when I don't need it to. I've tried outputting the results when there's a sound above a certain threshold but I can't eliminate the problem.

I was wondering, since I'm recording the incoming audio to a buffer if there's any objects or processes that can tell me the F0 in non-realtime?

non-realtime-f0-analysis

You probably want to try and remove the jittery-ness of the output of [pitch~]. Here's a patch that shows many different methods for smoothing data, have a look and perhaps implement one that works well for you.

thereishopeforus@hotmail.com wrote on Mon, 23 February 2009 12:39

Thanks for that, it is *very* interesting and useful. Will help a lot with things like Arduino input. Your patch has a great layout that lets you easily compare and understand the many ways of doing smoothing. Kudos!

thereishopeforus@hotmail.com wrote on Mon, 23 February 2009 12:39You probably want to try and remove the jittery-ness of the output of [pitch~]. Here's a patch that shows many different methods for smoothing data, have a look and perhaps implement one that works well for you.

Thanks for that, it is *very* interesting and useful. Will help a lot with things like Arduino input. Your patch has a great layout that lets you easily compare and understand the many ways of doing smoothing. Kudos!


I'd love to take credit for it but it's not mine, I found it here on the forum and I think it credits all the relevant people in the patch. The [lbyl] external from CNMAT might also be worth a look.

thereishopeforus@hotmail.com wrote on Mon, 23 February 2009 16:52

Speaking of that, I'm working on a better pitch tracker, based on a fast lifting wavelet transform. It's some months away yet, but it is much less jittery than other methods I've tried.

thereishopeforus@hotmail.com wrote on Mon, 23 February 2009 16:52I'd love to take credit for it but it's not mine, I found it here on the forum and I think it credits all the relevant people in the patch. The [lbyl] external from CNMAT might also be worth a look.

Speaking of that, I'm working on a better pitch tracker, based on a fast lifting wavelet transform.  It's some months away yet, but it is much less jittery than other methods I've tried.

Please consider doing a windows version of your new pitch tracker!

Thanks for these patches... going to come in very handy.

non-realtime F0 analysis