Sounds like: A drifting, half-tuned texture that sounds like a processed field recording.
A slowly evolving, semi-tonal texture that avoids sounding like a conventional 'synth pad' — closer to a processed environmental recording, with pitch and filter movement that never quite settles.
Open this patch in the interactive PatchpadALGO 8 with MIX at -64 selects the X output (Operator A modulating Operator C, plus B2 as a plain carrier), and A's ratio (3.14, non-integer) is what gives the semi-tonal, slightly inharmonic brightness -- not an abstracted 'FM texture' but a specific ratio choice. MOD page 1 and MOD page 2 are Digitone II's two independent LFOs -- running them at different, unsynced SPD values onto Filter Freq and Pitch respectively is exactly the two-independent-rates mechanism the brief wants. Corrected: MOD page 2 used to target 'Pitch' (an invented destination — Digitone II's FM TONE machine has no pitch destination for its LFOs). It now modulates RATIO A, the modulator's frequency ratio: at this small DEP it slowly shifts the inharmonic partial spacing, which is the closest FM TONE gets to 'pitch that never quite settles' (it moves the timbre, not the fundamental). For a true pitch-drift layer, WAVETONE's DRIF parameter is the native option on another machine.
Syntakt's melodic tracks use Digital machines, not Analog -- the earlier 'analog voice' idea doesn't apply here. SY SWARM's NMOD adds subtle noise modulation on top of the detuned oscillators, the closest this machine gets to a grainy, field-recording-adjacent texture without a literal sample-based voice. Melodic tracks 1-8 only offer Lowpass/Highpass/EQ on FLTR PAGE 1, not Bandpass (that's tracks 9-12 only), so Lowpass with high resonance stands in for the 'bandpass-like' emphasis. LFO PAGE 1 and LFO PAGE 2 -- Syntakt's audio tracks have two real LFOs, not one -- running at different, unsynced SPD values onto Filter Freq and Pitch give the two-independent-rates movement this patch wants. Corrected: this LFO used to target an invented 'Pitch' destination — Syntakt's LFO destination list (manual Appendix C) has no direct pitch option, so it now targets the machine's own TUNE knob (parameter 1), the real route to pitch modulation. (For hard one-shot pitch falls Syntakt also has native sweeps: BD MODERN's SWEP sets pitch-sweep depth and time.)
FILTER's SHAPE is set to a real BP (bandpass) rather than a plain lowpass, and its SOURCE is Mod Env 1 rather than Amp Env, giving the filter its own independent envelope. LFO 1 and LFO 2's destinations are hardwired on Summit -- filter cutoff and oscillator pitch respectively -- via their own section's Depth knob, so running them at different, unsynced RATE values realizes the 'two independent LFOs' brief without touching the Mod Matrix at all. LFO 2 DEPTH lives on the OSCILLATOR page by symmetry with LFO 1 DEPTH's documented home on FILTER; the manual doesn't spell this one out as explicitly, so treat it as inferred rather than manual-confirmed. A touch of MIXER's real Noise source is blended in for extra grain.
Osc 1 is Two Op. FM: Wave sets the ratio between the two operators (non-integer ratios are inharmonic and bell-like), Timbre is the modulation index and Shape is the feedback. LFO 1 runs a Tri wave at 2 Hz. LFO 2 runs a Tri wave at 0.4 Hz. Mod Matrix: the LFO 1 row to the Cutoff column at +25. Mod Matrix: the LFO 2 row to the Pitch 1+2 column at +10. All values here are on MiniFreak's native 0-100 scale, not 0-127.
The WAVE knob's 'VCO1 square FM'd by VCO2 triangle' position is a real, labeled Typhon setting that gives genuine inharmonic FM-style texture rather than an approximation. M1 and M2, both in LFO mode but running at different, unsynced RATE values onto CUT and CV (pitch) respectively, realize the two-independent-rates brief using two of Typhon's three real modulator slots -- M3 is left unused. Front-panel knobs are unmarked, so those values are clock positions rather than numbers; FILTER ADSR/AMP ADSR/M1/M2 use the OLED screen's real 0-100 values.
WaveScan held at a fixed mid-position (rather than swept) gives a semi-tonal, digitally-flavored texture. FILTER 1 uses the real BP Stn12 type and leans on Hydrasynth's pre-wired ENV 1 AMOUNT / LFO 1 AMOUNT connections -- Hydrasynth's version of a long filter envelope plus one LFO on cutoff, needing no Mod Matrix entry. LFO 2 is pre-wired to Amp by default, not pitch, so reaching Osc 1's pitch at a different, unsynced rate needs one explicit MOD MATRIX route, kept at low depth. Reverb is a real always-on module here, separate from Pre/Post-FX.
Using two LFOs at deliberately unrelated (non-synced) rates — rather than one — is what keeps this texture from settling into a predictable cycle, which is the whole point of the 'field recording' framing.