[sdiy] Dealing with velocity sensitivity / scaling on envelopes

[Joel B]

Ideally if the patch changes, all parameters should be eased (an easing function or a linear parameter fade) to the new sound,  the same if the voice is stolen - just morphs into the new sound - retriggering in that context would likely be annoying for pad and other sustained sounds, but for sounds with fast attacks might be best.

Joel
Sent from my iPhone 

> On Sep 4, 2017, at 8:08 AM, Mattias Rickardsson <mr at analogue.org> wrote:
> 
>> On 4 September 2017 at 13:44, Joel B <onephatcat at earthlink.net> wrote:
>> Wouldn't you just give each timbre its own fixed voice allocation and only steal within that timbre?
> 
> No, that makes the synth much more limited. (But I guess most makers would.)
> 
> Also, changing sounds (timbres, "presets") between notes is a case
> similar to voice stealing between timbres, even for a synth without
> either voice stealing or multitimbrality. A new sound is suddenly
> played with the voice - and what should happen then?
> 
> /mr
> 
> 
>> The old Kawai k5, Casio CZ, Cheetah MS6 work like that although the K5 has a VR mode where it will share voices between timbres, usually if I recall correctly the timbre changes and the envelope continues on as it was rather than re-trigger. The stealing effect is its own unique and rather cool sound imo.
>> 
>> Joel
>> 
>> 
>> 
>>>> On Sep 1, 2017, at 9:43 AM, Mattias Rickardsson <mr at analogue.org> wrote:
>>>> 
>>>>> On 1 September 2017 at 15:13, Oakley Sound <oakleysound at btinternet.com> wrote:
>>>>> 
>>>>> RTZ ADSRs have their place....
>>>> 
>>>> In the bin. :-)
>>> 
>>> Not if you're designing polyphonic multitimbral voice-stealing
>>> analogue synths! ;-)
>>> 
>>> This is a very tricky situation, and although I agree that the
>>> envelope should model a capacitor charging/discharging - and that the
>>> velocity factor should generally act as a multiplier on the resulting
>>> envelope rather than changing the envelope internally - and that the
>>> nasty clicks potentially introduced by RTZ can be hopelessly
>>> problematic, I clearly see the need of RTZ if there is a risk that the
>>> synth voice is doing something unknown and irrelevant before the note
>>> that is to be played. Starting the envelope from random values will
>>> ruin the sound to some extent, and this is what can happen if the
>>> voice is used for different timbres and/or sent to different outputs
>>> on consecutive notes.
>>> 
>>> Depending on your application, Tom, you might or might not need RTZ,
>>> but even without RTZ the velocity scaling could give you nasty
>>> clicks... so it's not easy! I'm even thinking that a zero-crossing
>>> detector would be useful in order to minimize those clicks when
>>> changing velocity, but that's perhaps the topic of another luxury
>>> discussion. :-)
>>> 
>>> /mr
>>> _______________________________________________
>>> Synth-diy mailing list
>>> Synth-diy at synth-diy.org
>>> http://synth-diy.org/mailman/listinfo/synth-diy