Re: Octal-dev digest, Vol 1 #123 - 5 msgs

[Jared]

AFAIK there aren't any phasing issues, as a multiply is just a scale, it's
when you start doing complex multiplications that the phasing starts.

A cooler way to do the cross-envelope would be as a simple vocoder type
thing (you know, that robot/keyboard/voice type thing). You get both input
signals and put them through a set of, say 8, filters that cover the whole
spectrum of the sound.  You then use the energy in each filtered B section
to change the amplitude of the sound in each filtered A section.  Then you
recombine A and spit it out again. You could even do whacky things like
making the high frequencies in A effected by the low frequencies in B etc.  
I saw this description on the net somewhere ages ago, with pretty pictures
and everything.

I to have only DSP experience in matlab (and a little bit of hacked DSP
assembler), and am really wanting to get into PC based stuff, so I am more
than willing to help out (and be helped out) on this one.  So Brendan, or
anyone else for that matter, give me a yell and lets do this sucker :)
J


>It looks like things are firming up in the code base.  Exciting!  I'd
>like to try writing some machines.
>   I have only done DSP type coding in matlab so my background is not
>great but I have a few ideas which I think could be simply implemented.
>The first, I would describe as a cross-envelope.  This machine would have
>two inputs, input A would be an audio signal which would be multiplied by
>the amplitude of input B.  The way I envision it being used is to create
>tempo patterns or humanizing factors.  For example, input A is a sampler
>playing a loop of some classical string tune.  Input B is a funky
>breakbeat from a drum machine.  The result is a sort of funky chopped
>sounding pizzacato version of the original loop.  There are other more
>bizzare and subtle uses but that was the one that came to mind
>first.  The same principle could be applied to filters or other effects,
>not just an amp.
>  So, the question for you signals geeks is, can I get away with just
>multiplying the amplitude of B by the input of A or will I have phasing
>problems?  Or is there an easier way to do it?

>rock on,

>-Brendan