Re: Pitch Shifting Question

[ Follow Ups ] [ Post Message ] [ The QuadraSynth Forum ] [ FAQ ]

Posted by sam on January 22, 1999 at 04:44:52:

In Reply to: Pitch Shifting Question posted by Sonny on January 15, 1999 at 21:53:56:

Sonny, that's a great question. I'll use the piano sample as an example here. The piano is a multi-sample, which means it has a new (looped) sample every few semitones. With more samples, the QS doesn't have to pitch transpose samples very far. Easy pitch transposition (like "speeding up the tape") gives a rather distinct coloration and artifacting (the chipmunk effect) to a harmonically rich sound. So more samples = less transposition and a nicer sound. The QS8 has a new piano sample every few semitones, typically 3-5.

Now for example's sake lets say you have a sample centered at D3 that is sped up or slowed to play the pitches from C3 to E3. If you tell the pitch unit to transpose the patch up by 2 semitones and play a D it speeds up the sample and sounds an E. Now you can set up the pitch envelope to slow the pitch back down, and you set an envelope to work like a square pulse (fast attack, high sustain, no sustain decay)

Then there are 2 controls that determine the modulation of the pitch envelope. The first is "penv->pitch (edit button 60, menu page 7) which sets the pitch envelope modulation amount (-99 to +99). This determines how the pitch envelope will pull the pitch up or down; it effectively multiplies the pitch envelope by a value from -1 to 1. If this multiplier is zero or the pitch envelope has no size (the next parameter) the pitch will not be transposed by the pitch envelope.

Then there's pitch envelope level (button 90 page 10) which sets the size of the envelope (values 0 to 99, and this is the value that gets modulated in the modmatrix). Important envelope concept: mod matrix modulators can ADD or SUBTRACT from this value from 0 to 99, and this final value will be multiplied by the penv->pitch value.

Continuing with the earlier example we could use the pitch envelope to pull the pitch down 2 semitones. You set the pitch envelope to some moderate negative value (ie -20 or something) and then increase the pitch env level until the envelope is big enough to pull the patch down just enough. I tune it using 2 voices in a patch with sine waves, then I'll revert back to the piano sample.

Now when you play a C3, it tries to pick the sample for D3 since you transposed the patch up, but then the pitch envelope slows the D sample back down 2 semitones, and you just played a note with the D3 sample pitched down to C, just as it would have been played if you hadn't done any of this. So what was the point?

Well if you play a B2, it picks the sample for C#3, which was that D3 sample, and it transposes it down rather than playing the A2 sample transposed up. What happens is that notes where the transposition makes you pick a different sample to transpose from will sound tonally different.

Under some circumstance (small transposition, no sample boundary crossing) it will not sound different at all, but small changes can make the insturment tend towards darker or brighter voicing (both with usefull and possibly annoying effects). If the transpostion forces you to a different sample for every note it can totally change the sound of the instrument.

So you can do this for interesting sounds and FX, and when combined with other sounds can be used to introduce or work around phase cancellation problems.

I hope this makes sense. BTW, all the envelopes work just like the one described here, and if you get the concept the layout of the envelope parameters in Unisyn is better than how the QS organizes these menus from the panel.


Follow Ups: