Sounds like: A lead with a fixed vowel-like 'aah' quality, slightly voice-like.
A lead with a fixed vowel-like resonant peak baked into the filter — not a true vocoder, but reads as slightly voice-like thanks to a narrow, resonant bandpass character.
Open this patch in the interactive PatchpadDigitone II's Multi-mode filter TYPE genuinely morphs LP/BP/HP -- setting TYPE to BP gives a real bandpass resonant peak rather than an LP-with-resonance approximation. ENV is kept low (10) since the vowel character comes from the static RESO peak, not an envelope sweep. FDBK acts on operator A in ALGO 1, so LEV (A) is raised to 50 on SYN page 2 to let the feedback colour carrier C; LEV (B) stays 0 and MIX -64 selects the C output only.
Syntakt's digital-voice tracks (1-8) only offer Lowpass/Highpass/EQ/OFF filter types -- no true Bandpass like Digitone II's own Multi-mode filter or Summit/Hydrasynth's dedicated BP mode -- so RESO pushed very high on a Lowpass is the closest available vowel-like resonant peak here. Melodic tracks use Digital machines, not Analog. AMP PAGE 2's MODE must be ADSR or SUS on AMP PAGE 1 is ignored.
Summit's FILTER SHAPE has a genuine BP mode -- a real bandpass peak, not a lowpass-with-resonance workaround. ENV DEPTH stays low (10) since the vowel comes from the static RESONANCE peak, not a sweep.
Osc 1 is Formant: Wave sets the interval between the two formants, Timbre the base formant frequency (the vowel position) and Shape the formant width. 0-100 scale throughout. Values are on MiniFreak's native 0-100 scale.
Typhon's own filter is a low-pass only. EG LEVEL is kept low since the vowel character comes from the static resonance peak, not an envelope sweep. Front-panel knobs are physically unmarked, described by clock position; AMP ADSR uses the OLED page's 0-100 convention. FX1 includes a resonant 2-pole high-pass filter (its CUTOFF, 20 Hz to 20 kHz, in series after the VCA), so Typhon's low-pass plus that high-pass in series forms a band-pass: close CUTOFF partway and raise FX1's high-pass to narrow the band around the vowel-like peak.
Filter 1's BP Stn12 is a real state-variable-style bandpass type -- a true resonant peak, not a lowpass-with-resonance workaround. ENV 1 AMOUNT stays low (10) since the vowel character is static, not swept.
A true bandpass filter mode (where available) gives a noticeably more convincing vowel character than lowpass-with-resonance.