[music-dsp] About volume ramping

[David Olofson]

On Saturday 01 April 2006 21:47, Didier Dambrin wrote:
[...]
> Everyone probably ramps volume a linear way, it works well and uses
> few CPU.  
> However, if CPU didn't matter, what would be the best, ideal method
> to ramp in the minimum time possible? Would this require filtering,
> as high freqs wouldn't require as much time to ramp as lower ones?

There is no single best way of doing it, because different 
applications have different requirements. Sometimes, getting a quick 
response is more important than minimizing distortion.

Anyway, if you really want minimum response times without audible 
distortion, you're probably going to have to make it frequency 
dependent one way or another. How about basically implementing a 
multiband compressor, but without the envelope tracking part?


> I'm concerned about ramping time because quick volume changes,
> needed for a synth's envelope, or fast LFO's, need the minimum time
> not to sound too smoothed up. But what's the ideal time?

Envelope control is different from smoothing volume control changes. A 
volume control is normally expected to just respond "reasonably 
fast", but without any audible artifacts whatsoever. Zipper noise in 
slow fades must be totally eliminated to make the volume control 
seriously usable.

Envelope control, OTOH, is part of the code DSP of the synth. That is, 
whatever it does becomes part of the synth's "soul", and the only 
thing that really matters is that it sounds appropriate for the type 
of synth your building/coding. Very fast envelopes in analog 
modelling synths are (generally) expected to, and *supposed* to 
click. If the sound programmer doesn't want clicks, he/she slows down 
the envelope. Pitch scaling of envelope times can come in handy here. 
(Obviously, this requires that fairly high parameter resolution for 
very short envelope times! Minor adjustments can be used to color 
desired clicks, so the accuracy must be pretty close to audio rate 
resolution for full control.)


> My method is to compute this ramping time based on the source &
> target level. Ramping from 0 to 100% takes 2x longer than from 50%
> to 100%.  
> However, this only works for normalized sources, and fails miserably
> when you think it's 50%, but it's actually much higher because of a
> post-amp later in the signal (and not predictable).

This doesn't make sense. Why would scaling down the chain make any 
difference to the quality of the ramping? (Unless you're talking 
about non-linear effects, that is.)


//David Olofson - Programmer, Composer, Open Source Advocate

.-------  http://olofson.net - Games, SDL examples  -------.
|        http://zeespace.net - 2.5D rendering engine       |
|       http://audiality.org - Music/audio engine          |
|     http://eel.olofson.net - Real time scripting         |
'--  http://www.reologica.se - Rheology instrumentation  --'