> > From what I understand it is impossible to get rid of aliased frequencies
> > after sampling.
> Once a frequency component is "aliased", it loses the original identity of
> that frequency takes on the identity of the alias. Then it's just a
> frequency component. The reason we hear these aliases is that, if the
> waveform was originally meant to be quasi-periodic (a.k.a. quasi-harmonic),
> the harmonics that are folded over cease to be harmonic and we hear that.
> So any harmonics that somehow find themselves higher in frequency than
> Nyquist will fold over and take on another frequency. How did they get
> there? The original recorded note just had them as harmonic. It is when
> notes that are rich in harmonics are pitched up to a higher pitch, that's
> when folding or aliasing may happen. Is that what is happening to you?
> I was trying to answer the question implied by the thread Subject header:
> "How does one generate bandlimited waveforms that can be morphed into other
> bandlimited waveforms?" Is that your question, André?
Yes, it is. I think I have enough information to try the waveform-tables
approach. I will quantise the morphing waveforms and do a FFT on each to build
wavetables. I looks like this is what I can accomplished with my current
Thanks for your input and insights. It is clear to me that all these problems -
so easy they look at the first glance - ends up really complicated if you want
to do the first thing that comes into your mind. Hope I can present the
solution for you to fiddle around with online soon.
dupswapdrop: music-dsp mailing list