The Voice source has a different value for each of the MiniFreak's 6 voices (or 12 in paraphonic mode).
This can be used to give each note a little bit different character. If you assign Voice to e.g. Attack time and set the amount to some value, and then play a chord, the attack time of each note will be slightly different which makes for a more lively sound. Likewise, playing notes in a row will cause a slight variation between them.
That is very useful to get the characteristics of old analog synth which had voices with slight differences in many of the parameters, such as envelope times, VCF resonance, the oscillator waveforms etc..
Jason Cooper has a website dedicated to this phenomenon and technique:
https://www.voicecomponentmodeling.com/You can read more the Keyboard sources in the manual "12. KEYBOARD FUNCTIONS".
