Hi guys,

My name is Ben Cole and I just downloaded puredata to use with a University project. I have viewed some beginner tutorials and the documentation, so I am beginning to understand how it works, but I still have some nooby questions.

Basically, I need to use puredata to analyze sound input from a microphone, and then send out the data over a port using the OSC protocol which the main program I am using, Virtools, can read. In PD, I need to analyze the best "beat per minute", frequency, and volume of the incoming audio - along with anything else I can figure out how to do.

I have no idea how to do any of this. I was hoping someone could point me in the right direction as far as what objects to use (adc~?) to input the live audio, break it down into the variables I need, and then send those out as OSC data.

ANY tips are appreciated.

Thanks,

Ben

use [bonk~] or any homemade alternative to recognise the transients, [alternate] to split the bangs into two alternating outputs, feed directly into [timer] to get millisecond data, then you need to do some maths to translate to bpm. - this is tricky though as this is assuming bonk gets only the crotchet (or quarter-note) waveforms and bonks it perfectly. a more sophisticated approach is to have the maths down so that your algorithm can work out when bonks are crotchets or otherwise, and ignore the occasional accidental-trigger to still work out the bpm. Logic(the software sequencer)'s bpm counter is very good at this

[fiddle~] to analyze frequency content of monophonic sources

and [env~] to gather amplitude data

pd-tutorial.com

To get you started, you can use [env~] to detect average amplitude. For pitch tracking, both [fiddle~] and [sigmund~] can be useful, though they'll take some playing around with to get good results for your input signal. According to the helpfile, you can actually get 3-voice polyphony from [fiddle~], but I've never tried and don't know how accurate it is. They can track amplitude as well. As for bpm, that can be a little tricky if you're just analyzing incoming audio. [bonk~] is useful for detecting transients, but I'm not really sure how to go about filtering out subdivisions. If you can, it might just be easier to make a tap tempo and follow along.

Hey guys - Thank you so much for your quick responses and valuable info!

I'll be trying them out in the next day or two and will post my findings.

As far as sending the data out as OSC, do I simply need to make an object [dumpOSC xxxx] (x's being the UDP port number) that has a variable feeding into it? I assume there is a bit more to it than that, but from what I understand I will be sending out a different piece of OSC data over a different port for each variable I will be using in virtools (IE - freq, bbm, amplitude). So far this link is all I really have relating to PD and OSC (http://www.aaronfryklund.com/wp/?p=45).

-Ben