[music-dsp] using the logarithmic pitch model in a sample playback engine ?

[James Chandler Jr]

I apologize in advance if this is a dumb idea, but perhaps you could create
your LFO so it has its normal wave-shape output, and it also has an
integrated LFO output.

That way the innards of the LFO could remain concerned with generating your
desired wave-shape, and the integrator would stretch the waveshape so it
would do the right thing when added to the delta value?

LFOIntegratedOutput += LFOOutput;

It would take a few processor cycles, but wouldn't require fancy math in the
loop. Setup calculations could be tricky, depending on how flexible you want
the LFO modulation to be. Ferinstance, it might get complicated if you want
to allow LFO's with a DC offset in the output.

This approach is simple, but it yields the correct FM shape and correct
pitch deviation when used to modulate the delta value in the circular buffer
of a chorus delay line. Perhaps the result would not be the same in a
sample-playback oscillator, but sample-playback pitch modulation appears to
use the same principle as pitch modulation in a circular delay line?

James Chandler Jr.

> I just used the "delta * (1 + lfo)" method myself, and it works fine
> for normal amounts of vibrato - but for very heavy vibrato not only
> is the shape distorted but more seriously the 'average' pitch drops.
>
> If the lfo wave is from a table I suppose it could be pre-distorted
> so the method worked perfectly (then you have to decide if the shape
> distortion is a problem when lfo depth modulation is used...)



dupswapdrop -- the music-dsp mailing list and website: subscription info,
FAQ, source code archive, list archive, book reviews, dsp links
http://shoko.calarts.edu/musicdsp/