Voice Modeling of Synth Envelopes
Exploring VCM on Envelope ADR Stages
Voice Component Modeling
You can read up more information on the general topic of Voice Component Modeling here:
http://www.VoiceComponentModeling.com
Per Voice Variance to Envelope Timing in Classic Synths
Classic poly synths are full of electrical components, including analog envelope generators (EGs) and associated circuitry. The intention and general methodology of developing poly synths has always been to create a great single voice architecture, and then duplicate that voice with multiple perfect copies, to allow multiple notes to be played at once for a given sound design. However, due to small differences in electrical tolerances between voices, and the aging of resistors, capacitors and other components over time, voices tend to not be exactly matched in performance. These imperfections lead to small variances in envelope ADR timing, on a per voice basis. The image to the right is an example of how four analog voices might differ at a given setting.
The audible effects of these ADR Envelope timing offsets are most pronouced with stacked, or unison sound designs, where multiple voices are triggered simultaneously and released simultaneously. In unison or stack patches where with high resonant filter amounts, the temporal offset of each voice's filter peak can produce a massive, organic filter sweep sound.
Modeling this Per Voice Classic Synth Behavior with Modern Synths
If your synthesizer offers Voice Number as a Modulation Source, or has an advanced mod sequencer like the Prophet Rev2, you can model per voice variance on your modern synth, giving your perfect digital envelopes a little less precision, and more character like classic analog envelopes when it comes to exact timing of ADR stages. Directly routing Voice Number mod source to Filter Attack, Amp Attack, or both, will create slight timing offsets to their performance. Note that modulating Attack has a direct downstream affect on the Decay variance per voice as well. For the most detailed voice modeling of envelopes, you can target Attack, Decay and Release with voice number modulation to give each voice slight variance to its ADR stages. You can play around with the amounts to dial in the amount of character you prefer. As mentioned above, the effect is most pronounced in unison mode with multiple voices stacked together with simultaneous attack and release impulses, so the best way to test the amounts would be in unison mode - to get an idea of how much variance / timing separation there is from voice to voice. For synths that don't have Voice Number mod source, or a sufficiently advanced mod sequencer, you can also try using an LFO at a very slow rate, set to random with key sync on, to give a more randomized amount of variance per voice.
You can read more about Voice Modeling Techniques on the Voice Component Modeling Website.
Envelope Contours and Recursive Envelope Shaping
The specific curves/contours of the Attack, Decay and Release stages vary from synth to synth, however that is a separate topic. Most synths have contours similar to the image shown above, however some synths have had linear envelope progression as well (Juno). For more info on Recursive Envelope Shaping, click here.