additive sythesis theoretical failures...


    Feb 03 2006 | 12:56 am
    I have been working on a demo of additive synthesis for a class. I wanted to show the construction of tri, square, and sawtooth waves as mere sums of harmonics of sine waves. So i made a poly with a simple patch that mutes even harmonics for triangle and square, and adjusts the amplitudes (1/harm#^2 for tri, 1/harm# for sq. and saw) as necessary. The thing is, the oscilliscope doesn't show a waveform that looks anything like it is supposed to (though the spectral content looks about right). So I guess I need to make some sort of phase adjustment so the picture looks right... How can I show the class that science is TRUE? :)
    and, in the spirit of things, a Wendy Carlos instructional sample: http://www.yowstar.com/share/sinesqorwhite.mp3
    peter
    on a related note, why aren't the anti-aliased oscillators normalized from -1 to 1?

    • Feb 03 2006 | 8:13 am
      > I have been working on a demo of additive synthesis for a class. I wanted to show the construction of tri, square, and sawtooth waves as mere sums of harmonics of sine waves. So i made a poly with a simple patch that mutes even harmonics for triangle and square, and adjusts the amplitudes (1/harm#^2 for tri, 1/harm# for sq. and saw) as necessary. The thing is, the oscilliscope doesn't show a waveform that looks anything like it is supposed to (though the spectral content looks about right).
      Looks okay to me, mate - you just need to scale the waveform down some.. Remember that unless you set the frequency high you're only getting a small fraction of the necessary partials.
      > on a related note, why aren't the anti-aliased oscillators normalized from -1 to 1?
      It's not possible to normalize them exactly because the peak value can't be predicted. Something seemed wrong about scaling to the same amplitude as cycle~ and having the signal "clip" above 1, so I went with the half amplitude instead... like phasor, I guess.
      Ben