Hi,

I'm interested in using PureData to analyze incoming audio (from a DJ) to control videos in GEM. The basic idea is to create a VJ style patch that acts semi-autonomously by extracting tempo and frequency data to manipulate video playback (changing framerate, RGB, fade between videos, etc.).

I've started using pd-aubio library and it seems to be pretty good at getting out tempo, and onsets (if I [bp~] to find particular areas of interested in the track) . It's less good at getting out accurate pitch data but I expect this to be more difficult when I'm just analyzing a stereo signal coming from a soundfile or from the [adc~].

I'd like to ask if there are any useful methods and tricks people are aware of that I could implement within my patch. I'm new to this sort of thing and I'm essentially fumbling around in the dark at the moment. Are there methods any of you have come across that might be handy for this sort of thing?

Cheers,
Chris

There are several object for pitch detection. [sigmund~] could be useful to you. Katja Vetter has coded a [helmholtz~] pitch trackig object which is also good. [fiddle~] can also do that (but I think it has more or less been replaced by [sigmund~]).
If that's not enough, then you should probably consider using FFT to analyze the incoming signal.

Hi @alexandros, thanks for your post!

I've become aware of these objects from doing some research on the forums but I haven't really figured out how best to implement them. One main problem is that I'm using a polyphonic stereo signal with lots of things going on (bass, drums, synth lines, etc.), so pitch estimation objects like sigmund~ and fiddle~ seem to give pretty spurious results at times. Admittedly, I am using them in a very basic way. I would be interested in seeing examples of more complex ways I could use these for analysis and extraction of information.

I've had a quick glance over the helmholtz~ documentation and it seems like it could be a pretty powerful tool to use. Again, it seems better suited to monophonic rather than polyphonic signals though. If you have any further ideas of how I could get around this problem, I would be open to such suggestions.