I'm attmepting to make a patch that will generate visuals based on the properties of a sound file e.g. bass treble etc. For this I need to be able to read or get the values of sound file.

What I think I need is a graphic equaliser. I have modified the patch found here and whilst it works is this the right way to do it?

http://www.pdpatchrepo.info/hurleur/eq.pd

Well, I have a gut feeling that that eq is not the best you can do - how a does a sound file sound when it's set flat?

In a sense, who cares, right? By taking the env~ value, you're getting information that is related to amplitude at a particular frequency band and using one of the cheapest filters to get it. I don't know anything about graphics but you may want to set the env~ refresh rate at a similar rate to your graphics processing.

You can really get crazy with this you know, it doesn't really matter where you get the numbers from, just experiment with different analysis techniques. And can't you have the audio signal operate directly on a graphic?

Someone want to suggest FFTs?

And then I imagine that peak~ or anything that can detect attacks in drum beats accurately would be indispensable for a visualizer.

Search this forum -

And can't you have the audio signal operate directly on a graphic?

How would one do this?

The general idea is to create a series of cubes in GEM and have properties of the sound affect the shape, size, colour, rotation and other parameters of the cube. For example, the bass of the sound will make the cube bigger.

I'll look into fft and peak~ but, to be honest, I'm more adept with visuals than I am with sound

It doesn't matter - all you need are numbers. Experiment - perhaps you don't get the results you want, but consider that you are doing more than 99% of VJs that are just mixing pre-made videos. Only by experimenting and getting some info from the audio to the visual will you learn how to design predictable results. Don't just load up fft~, start with the patch you posted and connect the output of the lowest env~ to the size parameter of your cubes and see what happens! Adjust maths until it works. Out of convenience you can [unsig~] audio signals, but again, someone else needs to speak to that.

In my simple mind I imagine simply displaying (and then altering) the signal waveform in GEM - betting there's an object for this, have you seen:

http://en.flossmanuals.net/pure-data/ch086_gem/

Do you actually need to edit the sound itself? I think using [bp~] and [env~] are probably fine for a basic analysis like this. But I wouldn't sum their outputs back up and feed it to the [dac~]; just send the original sound. Graphic equalizers are pretty ugly sounding.

Thanks for all of your help with this so far!

Maelstorm, you were right about the output sounding crappy. I want to mix in generative visuals with things that I can control so I'll keep it in so that I can compose visuals to the music in realtime.

I've attached the patch that I've got so far. Load a wav file then press start. Changing the input knob will make the visuals react more, though you can amplify each channel separately using the white sliders.

http://www.pdpatchrepo.info/hurleur/eq.pd

Yeah Maelstorm is right - best case scenario is that you are receiving an aux output from a hardware mixer.

Katja just posted in index/pixel regarding an object called [scopeXYZ~] it sounds like it will display a waveform for you.

Thanks for [scopeXYZ], I might just use it as well, though it looks more useful for these kinds of visuals. For those not in the know I've attached a patch demonstrating [scopeXYZ~]

http://www.pdpatchrepo.info/hurleur/scope_test.pd