Does this look right for phase modulation of live input?
[adc~] [0\
| |
[rfft~] [osc~]
| \ |
| \ | [0\
| \ | |
[cartopol~] [*~]
| | |
| [+~]
| |
[poltocar~]
| /
| /
| /
[rifft~]
|
[dac~]
-
Live phase modulation
-
You can just do it with a delay line:
[adc~]
|
[delwrite~][osc~]
|
[*~]
|
[vd~]Remember, the phase of an oscillator is just a table lookup of a waveform. Using a delay line, it's just a table lookup of continuously recorded material.
-
I'm using this technique as live frequency modulation, but comparing the classic FM and PM techniques with two oscillators, you get quite different sound. Converting from Cartesian to Polar after [rfft~], you get the phase out of the right outlet, don't you? So adding an oscillator there should be messing around with the phase...at least that's what I had in mind. Plus these two techniques (the one I posted and the you did) do sound different.
I mean, is there a point going on with this, trying to make it a bit better? -
the 'phase' in fft is quite different to the phase in normal signals.
it's an interesting idea though....does it sound interesting?
-
It actually does. If I apply windowing and overlapping it might be improved. I'll do that in the next few days.
-
Yeah, the example I gave was a little over-simplified. If you want to have more predictable results, you have to do some scaling of the modulator based on the index of modulation and the carrier frequency. I've attached a more thorough example which plots the standard two-oscillator approach and the delay-based version for comparison. If you are using an arbitrary input and want to maintain the C:M ratio, you'll obviously need to modify this to use a pitch-tracker.
If you're interested in the research, there's a paper on it here:
-
Nice! Thanks a lot. The math is a bit over my head, but I'm taking classes, so I'll get there.
Cheers
-
Ok, coming back to it after comparing my attempt with Maelstorm's patch. I did windowing and overlapping to the patch in the first post of this thread but it sure doesn't sound as good as Maelstorm's with a pitch tracker included (with an input from my guitar). Just to let anyone know. who sees this post and finds the idea interesting.
-
The FFT approach is very different. With a wavetable approach, you are modulating the phase of one single waveform. Even if you have many frequencies in that waveform, you are modulating them all as a single wave. You're using units of samples, and how much of the phase a sample is is frequency-dependent.
With FFT, you are modulating them all separately. And you are modulating them in radians, which are frequency independent. pi/2 radians (or 90 degrees) is a quarter of a sinewave, no matter what the frequency of the sinewave is. 50 samples is pi/2 radians at 220.5 Hz, but it's pi radians at 441 Hz.
