Hey guys in this oportunity I would like to create a Shepard Tone in PD, I need to create it with 3 oscilators separated by an Octave. You know this experiment is a Cromatic Scale that is going up by semi tones until finish in the other octave, the percepcion is hear that this scale going up infinitively.

Ok We need to create three scales, and we have to change the velocity for each one, the first one (higher) is from the strongest until softest, second one (middle) It's velocity doesn't change, is constantly. Finally latest one (Lowest) is from the softest until the strongest.

Could you help me to create one ?

@ManoloPureD Hello......
Look in your Pd installation folder under pd/doc/3.audio.examples/ for the file "D09.shepard.tone.pd".
That will get you started........ and probably finished too......

There was also a folder in Pd extended that might interest you, that has been lost in Vanilla (so much good work has gone missing).
It is a few patches that demonstrate the anomalies of the human perception of audio.
It might well work in Vanilla though......... Cognition.zip
David.

@whale-av Hey thanks bro, I had not idea about this Patches in Pd intallation folder, the file "D09.shepard.tone.pd" is ok, that would be a good reference. But I need to create one in form of a scale not at Glissando, I mean the user should hear a scale with intervals per semi tone. How can I make that ?

I don't know why, but i made this: shepardtone.pd. The display can also be used to change the pitch.

@ingox Awesome man, It's perfect. My only problem is that I need to explain that, Could you help me to explain How do you create that ? It's absolutely good man, thanks!

@ManoloPureD Well, there is some stuff going on with data structures and some tricks with a hidden hslider to get an infinite slider. But the main stuff for the audio is this:

Here you have a continuous input from minus infinity to infinity. The [mod 384] makes it circular between 0 and 383. But we divide by 384 so the input range doesn't matter, it only defines the resolution. The output here is between 0 and 1.

From here the actual stuff happens within the expressions. We want the pitch to double for each octave, so we have pow(2, $f1). So now we have a lowest pitch at 0 and the highest pitch at 1. Playing with factor 4 and 110 can give different results.

Now we want the volume to be at maximum in the middle and at zero at 0 and 1 so we can not hear that the pitch switches. This is a little tricky and done by using the absolute function. I think the power function is only there to smooth things out.

I actually had to play around myself quite a bit to get the result, which can still be optimized, i guess. So i would recommend to just deconstruct the expressions and play around with some values yourself, using print, to get a feeling for what is going on.

@ingox Ok man I will try to change some things, in general is not complicated at all I guess. Thanks man. Wish me luck, I'll be bothering you later!

@ingox said:

I don't know why, but i made this: shepardtone.pd. The display can also be used to change the pitch.

Great patch, I like the data structure visuals and the shepard sound.. Thanks for sharing.
Alberto