Parameter reference

Every parameter, by the page it appears on

The ID under each name is what presets and mse-render use (power = 9). Ranges and defaults are read from the plug-in’s source when this site is built. A parameter can appear on more than one page (the Main page repeats the essentials).

A level of −60 dB means off for level controls, and −100 dB means off for gates and hiss.

Main page

Parameter	Range	Default	What it does
Sync sync_mode	Free · Tempo	Tempo	Tempo: trace length follows the host’s tempo and traces line up with its bar grid. Free: trace length is set in milliseconds and runs from when playback starts.
Trace Length (Sync) trace_division	1/16 · 1/8 · 1/4 · 1/2 · 1 bar · 2 bars · 4 bars	1 bar	Trace length when Sync is Tempo: 1/16 note to 4 bars. The default is 1 bar.
Trace Length (Free) trace_ms	10 – 20000 ms	2000 ms	Trace length when Sync is Free: 10 ms (grains) to 20 s.
Memory Capacity capacity not automatable	1 – 4000 traces	100 traces	The most traces memory holds (1–4000). Changing it keeps the traces that fit: clamped ones first, then the newest.
Activation Power power	1 – 9	3	The activation exponent. 1 = every memory contributes (a blurred schema); 9 = only the closest memories answer.
Cue Mode cue_mode	Segment · Progressive · Rolling	Segment	Segment: each finished segment cues memory for the next (a delay-like echo). Progressive: memory is re-cued as each segment unfolds. Rolling: no grid; the last Rolling Window of input is searched for inside every trace.
Mode Selector mode_selector	Custom · 1 · 2 · 3 · 2+3 · 1+2 · 1+3 · 1+2+3 · Iterative 1+2+3	Custom	The RE-201’s head combinations: 1, 2, 3, 2+3, 1+2, 1+3, 1+2+3 (delay heads), Iterative 1+2+3, or Custom (use the per-head settings below).
Blend Domain blend_domain	Waveform · Spectral	Waveform	Waveform: the echo sums trace audio (exact, but unaligned memories can cancel). Spectral: magnitudes are summed and each frequency takes the phase of the memory loudest there, so memories blend without cancelling.
Sequence Context sequence_context	Off / On	Off	Every trace also stores the previous segment’s address, [n−1 | n]. On: cues use that context (Context Cue). Context is recorded either way, so you can switch it on later.
Address address	Spectrum · Pitch Class · Pitch · Timbre · Rhythm · Custom	Spectrum	Which address memory compares (see Addresses): Spectrum (brightness and envelope, the original), Pitch Class (notes, octave-blind), Pitch (notes with register), Timbre (the sound, not the note), Rhythm (where the attacks fall), or Custom (a weighted mix, set with the weights below). Every trace stores all five, so you can switch at any time.
Freeze Memory freeze	Off / On	Off	Stop storing and stop forgetting. Memory keeps answering, like the Space Echo’s infinite repeat.
Feedback feedback	0 – 1.2	0	Echo mixed back into what gets recorded (soft-clipped; above 1 is allowed).
Echo Level echo_level_db	-60 – 6 dB	0 dB	Echo level. −60 dB = off.
Dry Level dry_level_db	-60 – 6 dB	0 dB	Dry level. −60 dB = off.

Memory page

Parameter	Range	Default	What it does
Memory Capacity capacity not automatable	1 – 4000 traces	100 traces	The most traces memory holds (1–4000). Changing it keeps the traces that fit: clamped ones first, then the newest.
Memory Budget memory_budget not automatable	128 MB · 256 MB · 512 MB · 1 GB · 2 GB · 4 GB	1 GB	Audio memory reserved for traces (128 MB – 4 GB). With many traces the budget limits the longest trace a slot can hold; the status bar shows that limit (slot max).
When Full full_policy	Oldest · Random · Least Used · Weakest · Merge Similar · Reject	Oldest	Which trace makes room when memory is full: Oldest (tape-like), Random, Least Used (lowest recent activation), Weakest (most faded), Merge Similar (average the new trace into its closest unclamped one) or Reject (keep what’s there). Clamped traces are never replaced.
Consolidate Above merge_threshold	0 – 1	1	Merge a new trace into an existing one whenever they are at least this similar, even if there is room. Prototypes form. 1 = off.
Freeze Memory freeze	Off / On	Off	Stop storing and stop forgetting. Memory keeps answering, like the Space Echo’s infinite repeat.
Write Mode write_mode	Auto · Manual	Auto	Auto stores every trace that passes the gates. Manual stores only captured traces.
Write Gate write_gate_db	-100 – 0 dB	-70 dB	Traces quieter than this aren’t stored. −100 dB = off.
Novelty Gate novelty_mode	Off · Store Novel · Store Familiar	Off	Store Novel: store a trace only if nothing in memory is this similar. Store Familiar: only if something is. Off: no novelty test.
Novelty Threshold novelty_threshold	0 – 1	0.9	The similarity used by the Novelty Gate.
Write Probability write_probability	0 – 1	1	The chance that a trace is stored.
Record Source record_source	Input · Echo · Input + Echo	Input	What gets stored: the Input (plus feedback), the Echo (memory feeds on its own output and keeps sustaining after the input stops), or both.
Clamp Incoming clamp_incoming	Off / On	Off	Clamp every newly stored trace (within the clamp budget).
Clamp Budget clamp_budget	0 – 1	0.5	The largest fraction of memory that may be clamped, so memory can still change.
Clamped Don’t Decay clamp_protects	Off / On	On	Clamped traces are exempt from forgetting (fading and feature loss).

Forgetting page

Parameter	Range	Default	What it does
Encoding Failure (Lf) encoding_failure	0 – 1	0	MINERVA’s L: the probability that each feature of a stored trace’s address is lost at encoding. The heard cue stays intact.
Tape Dropouts content_dropout	0 – 1	0	At recording, each 1/32 of a trace may drop out (a tape dropout).
Forgetting / Segment decay_forget	0 – 0.2	0	Each segment, every unclamped trace loses each of its features with this probability: recall gets vaguer.
Fading / Segment decay_fade_db	0 – 6 dB	0 dB	Each segment, every unclamped trace loses this much strength (activation). Traces below −60 dB are forgotten.
Wear Tone wear_tone	0 – 1	0	Older traces play back duller, like worn tape.

Address page

Parameter	Range	Default	What it does
Address address	Spectrum · Pitch Class · Pitch · Timbre · Rhythm · Custom	Spectrum	Which address memory compares (see Addresses): Spectrum (brightness and envelope, the original), Pitch Class (notes, octave-blind), Pitch (notes with register), Timbre (the sound, not the note), Rhythm (where the attacks fall), or Custom (a weighted mix, set with the weights below). Every trace stores all five, so you can switch at any time.
Spectrum Weight spectrum_weight	0 – 1	1	Custom address: how much the Spectrum set counts.
Pitch Class Weight pitch_class_weight	0 – 1	0	Custom address: how much the Pitch Class set counts.
Pitch Weight pitch_weight	0 – 1	0	Custom address: how much the Pitch set counts.
Timbre Weight timbre_weight	0 – 1	0	Custom address: how much the Timbre set counts.
Rhythm Weight rhythm_weight	0 – 1	0	Custom address: how much the Rhythm set counts. Similarity is the weighted mean of the sets’ similarities; with every weight at 0 the Spectrum is used.
Feature Focus feature_focus	Full · Rhythm · Timbre	Full	Within the Spectrum set: compare it Full, Rhythm only (loudness over time) or Timbre only (average spectrum). Segment and Progressive cueing. The Address choice offers finer Rhythm and Timbre sets.
Features feature_mode	Continuous · Ternary	Continuous	Continuous feature values, or classic MINERVA −1/0/+1 (Ternary).
Ternary Threshold ternary_threshold	0 – 2	0.5	Ternary: how far (in standard deviations) a feature must be from the trace’s average to count as +1 or −1.
Similarity similarity	Hintzman · Cosine	Hintzman	Hintzman (MINERVA II’s normalised dot product) or Cosine.
Activation Power power	1 – 9	3	The activation exponent. 1 = every memory contributes (a blurred schema); 9 = only the closest memories answer.

Retrieval page

Parameter	Range	Default	What it does
Activation Power power	1 – 9	3	The activation exponent. 1 = every memory contributes (a blurred schema); 9 = only the closest memories answer.
Similarity similarity	Hintzman · Cosine	Hintzman	Hintzman (MINERVA II’s normalised dot product) or Cosine.
Address address	Spectrum · Pitch Class · Pitch · Timbre · Rhythm · Custom	Spectrum	Which address memory compares (see Addresses): Spectrum (brightness and envelope, the original), Pitch Class (notes, octave-blind), Pitch (notes with register), Timbre (the sound, not the note), Rhythm (where the attacks fall), or Custom (a weighted mix, set with the weights below). Every trace stores all five, so you can switch at any time.
Features feature_mode	Continuous · Ternary	Continuous	Continuous feature values, or classic MINERVA −1/0/+1 (Ternary).
Ternary Threshold ternary_threshold	0 – 2	0.5	Ternary: how far (in standard deviations) a feature must be from the trace’s average to count as +1 or −1.
Self Match self_match	Off / On	On	On: the segment just played can answer its own cue (the delay-like part). Off: memory answers only with earlier traces.
Cue Gate cue_gate_db	-100 – 0 dB	-60 dB	Segments quieter than this don’t cue memory, so the noise floor doesn’t trigger echoes.
Negative Activations negative_mode	Subtract · Ignore · Absolute	Subtract	What negative activations do: Subtract (phase-inverted, true MINERVA), Ignore, or Absolute.
Echo Normalization normalization	Sum · Max · Familiarity	Sum	Sum (constant loudness), Max, or Familiarity (the echo is quiet when the cue resembles nothing in memory).
Echo Level Tracking level_tracking	0 – 1	1	1: the echo’s level follows the cue’s, like a tape repeat, so feedback decays. 0: memories return at their own level.
Recency recency	0 – 1	0	Favour recently stored traces (0 = no preference).
Feature Focus feature_focus	Full · Rhythm · Timbre	Full	Within the Spectrum set: compare it Full, Rhythm only (loudness over time) or Timbre only (average spectrum). Segment and Progressive cueing. The Address choice offers finer Rhythm and Timbre sets.
Cue Noise cue_noise	0 – 1	0	Random noise (in standard deviations of the address) added to every cue: live, chain, iterative and progressive. Knocks retrieval out of ruts; also fuzzy hearing for live cues.
Habituation habituation	0 – 1	0	Traces that just answered are less active for a few segments (their fatigue halves each segment), so memory moves on instead of repeating itself.

Cueing page

Parameter	Range	Default	What it does
Cue Mode cue_mode	Segment · Progressive · Rolling	Segment	Segment: each finished segment cues memory for the next (a delay-like echo). Progressive: memory is re-cued as each segment unfolds. Rolling: no grid; the last Rolling Window of input is searched for inside every trace.
Cue Source cue_source	Input · Sidechain · Random · Frozen · Echo Chain	Input	What cues memory: the Input, the Sidechain (another track routed to the plug-in’s sidechain input), a Random address (memory dreams), Frozen (the last live cue, held), or Echo Chain (each echo cues the next; see Sequences).
Progressive Start progressive_start	1 – 15 slots	2 slots	The slot (of 16 per segment) from which the progressive cue takes over.
Rolling Window rolling_window_ms	60 – 2000 ms	250 ms	Rolling: the length of the live cue.
Rolling Interval rolling_interval_ms	20 – 1000 ms	100 ms	Rolling: how often memory is searched.
Prediction lookahead_ms	0 – 2000 ms	0 ms	Prediction: play memory this far ahead of the current position, so you hear what came next last time.
Cue Smoothing cue_smoothing_ms	1 – 500 ms	40 ms	Crossfade whenever a live cue (Progressive or Rolling) changes the echo.

Sequence page

Parameter	Range	Default	What it does
Sequence Context sequence_context	Off / On	Off	Every trace also stores the previous segment’s address, [n−1 | n]. On: cues use that context (Context Cue). Context is recorded either way, so you can switch it on later.
Context Cue context_cue	Match Both · Predict Next · Current Only	Predict Next	Match Both: the last two segments against [n−1 | n] (sequence-sensitive recall). Predict Next: the last segment against traces’ n−1 halves, so you hear what came next last time. Current Only: ordinary MINERVA II.
Context Weight context_weight	0 – 1	1	Match Both: how much the n−1 half counts (0 = like Current Only, 1 = as much as the current half).
Cue Source cue_source	Input · Sidechain · Random · Frozen · Echo Chain	Input	What cues memory: the Input, the Sidechain (another track routed to the plug-in’s sidechain input), a Random address (memory dreams), Frozen (the last live cue, held), or Echo Chain (each echo cues the next; see Sequences).
Chain Input chain_input	0 – 1	0	Echo Chain: how much the live input steers the chain. 0 = memory walks on its own; 1 = the same as Predict Next from the input.
Chain Step chain_step	Blend · Sample	Sample	Echo Chain: what cues the next step. Sample: one of the answering traces, drawn in proportion to its activation (a random walk through memory; with Playback = Sample it is the trace you hear). Blend: the echo’s blended content (deterministic; tends to settle on a prototype).
Address address	Spectrum · Pitch Class · Pitch · Timbre · Rhythm · Custom	Spectrum	Which address memory compares (see Addresses): Spectrum (brightness and envelope, the original), Pitch Class (notes, octave-blind), Pitch (notes with register), Timbre (the sound, not the note), Rhythm (where the attacks fall), or Custom (a weighted mix, set with the weights below). Every trace stores all five, so you can switch at any time.
Cue Noise cue_noise	0 – 1	0	Random noise (in standard deviations of the address) added to every cue: live, chain, iterative and progressive. Knocks retrieval out of ruts; also fuzzy hearing for live cues.
Habituation habituation	0 – 1	0	Traces that just answered are less active for a few segments (their fatigue halves each segment), so memory moves on instead of repeating itself.
Activation Power power	1 – 9	3	The activation exponent. 1 = every memory contributes (a blurred schema); 9 = only the closest memories answer.
Self Match self_match	Off / On	On	On: the segment just played can answer its own cue (the delay-like part). Off: memory answers only with earlier traces.
Playback playback	Blend · Voices · Sample	Blend	Blend: every active memory, weighted (MINERVA’s echo). Voices: the strongest few as separate chorus voices. Sample: one memory at a time, drawn at random by activation.
Cue Mode cue_mode	Segment · Progressive · Rolling	Segment	Segment: each finished segment cues memory for the next (a delay-like echo). Progressive: memory is re-cued as each segment unfolds. Rolling: no grid; the last Rolling Window of input is searched for inside every trace.

Heads page

Parameter	Range	Default	What it does
Mode Selector mode_selector	Custom · 1 · 2 · 3 · 2+3 · 1+2 · 1+3 · 1+2+3 · Iterative 1+2+3	Custom	The RE-201’s head combinations: 1, 2, 3, 2+3, 1+2, 1+3, 1+2+3 (delay heads), Iterative 1+2+3, or Custom (use the per-head settings below).
Head 1 Level head1_level_db	-60 – 6 dB	0 dB	Level of head 1, the main echo.
Head 1 Pan head1_pan	-1 – 1	0	Pan of head 1.
Head 2 head2_mode	Off · Delay · Iterative	Off	Custom mode only. Delay: replays what head 1 played one segment earlier. Iterative: cued by head 1’s echo (the echo of the echo).
Head 2 Level head2_level_db	-60 – 6 dB	-6 dB	Level of head 2.
Head 2 Pan head2_pan	-1 – 1	-0.5	Pan of head 2.
Head 3 head3_mode	Off · Delay · Iterative	Off	Custom mode only. Delay: replays what head 1 played two segments earlier. Iterative: cued by head 2’s echo.
Head 3 Level head3_level_db	-60 – 6 dB	-9 dB	Level of head 3.
Head 3 Pan head3_pan	-1 – 1	0.5	Pan of head 3.
Playback playback	Blend · Voices · Sample	Blend	Blend: every active memory, weighted (MINERVA’s echo). Voices: the strongest few as separate chorus voices. Sample: one memory at a time, drawn at random by activation.
Voices voices	1 – 8	4	Voices: how many memories play.
Voice Spread voice_spread	0 – 1	0.7	Voices: how far they fan out across the stereo field (the strongest stays in the centre).
Voice Detune voice_detune	0 – 50 cents	8 cents	Voices: detuning of the outermost voices.
Voice Delay voice_delay_ms	0 – 50 ms	12 ms	Voices: delay of the outermost voices.

Tape page

Parameter	Range	Default	What it does
Length Mismatch length_mismatch	Varispeed · Cut · Loop · Stretch	Varispeed	A trace recorded at another trace length (you changed the repeat rate or the tempo): Varispeed plays it faster or slower to fit (pitch follows, like tape), Cut plays it as recorded, Loop repeats a shorter trace, Stretch time-stretches it to fit and keeps its pitch.
Wow wow	0 – 1	0	Slow, drifting speed wobble (~0.5 Hz, up to 6 ms).
Flutter flutter	0 – 1	0	Fast speed wobble (~7 Hz, up to 0.5 ms).
Tape Drive tape_drive	0 – 1	0	Soft saturation of the echo (unity for quiet signals); also shapes what feeds back.
Hiss hiss_db	-100 – -30 dB	-100 dB (off)	Tape noise on the echo. −100 dB = off. With feedback it gets recorded too.
Feedback Bass feedback_bass_db	-12 – 12 dB	0 dB	Low shelf (250 Hz) inside the feedback path: each repeat gets fuller or thinner.
Feedback Treble feedback_treble_db	-12 – 12 dB	0 dB	High shelf (3 kHz) inside the feedback path: each repeat gets brighter or darker.
Spring Level spring_level_db	-60 – 6 dB	-60 dB (off)	Spring reverb on the echo. −60 dB = off.
Spring Decay spring_decay	0.3 – 6 s	2 s	Spring reverb decay time.
Spring On Dry spring_on_dry	Off / On	Off	The spring also gets the dry signal, as in the RE-201’s reverb modes.
Echo Tone echo_tone_hz	200 – 20000 Hz	20000 Hz	A low-pass on the echo. 20 kHz = off.
Familiarity > Tone intensity_to_tone	-1 – 1	0	Familiarity (the strongest activation) moves the echo tone: familiar echoes brighter (+) or darker (−).
Familiarity > Feedback intensity_to_feedback	-1 – 1	0	Familiarity changes the feedback: familiar echoes feed back more (+) or less (−).

Spectral page

Parameter	Range	Default	What it does
Blend Domain blend_domain	Waveform · Spectral	Waveform	Waveform: the echo sums trace audio (exact, but unaligned memories can cancel). Spectral: magnitudes are summed and each frequency takes the phase of the memory loudest there, so memories blend without cancelling.
Spectral Voices spectral_voices	1 – 32	12	Spectral: how many of the most active memories are analysed each frame (1–32).
Spectral Freeze spectral_freeze	Off / On	Off	Hold the echo’s spectrum as a drone. Memory can change or be cleared underneath it.
Max Active Traces max_active	1 – 512	64	A blend uses at most this many of the most active traces (the level is kept). Keeps huge memories cheap.
Length Mismatch length_mismatch	Varispeed · Cut · Loop · Stretch	Varispeed	A trace recorded at another trace length (you changed the repeat rate or the tempo): Varispeed plays it faster or slower to fit (pitch follows, like tape), Cut plays it as recorded, Loop repeats a shorter trace, Stretch time-stretches it to fit and keeps its pitch.

Mix page

Parameter	Range	Default	What it does
Feedback feedback	0 – 1.2	0	Echo mixed back into what gets recorded (soft-clipped; above 1 is allowed).
Echo Level echo_level_db	-60 – 6 dB	0 dB	Echo level. −60 dB = off.
Dry Level dry_level_db	-60 – 6 dB	0 dB	Dry level. −60 dB = off.
Edge Fade edge_fade_ms	0 – 50 ms	5 ms	Fade length at trace edges and when the echo changes (declicking).
Output Gain output_gain_db	-60 – 12 dB	0 dB	Output gain.

Control page

Parameter	Range	Default	What it does
Running running	Off / On	Running	Paused: memory is left exactly as it is (nothing stored, forgotten or cued), the echo fades out and the dry signal passes. Resuming starts cleanly at the next segment. Not changed by presets.
MIDI Note Control midi_control not automatable	Off / On	On	MIDI notes trigger actions (see MIDI and automation).
MIDI Channel midi_channel not automatable	0 – 16	0	The channel MIDI notes are read on (0 = any).
MIDI Base Note midi_base_note not automatable	0 – 100	36	The first note of the MIDI map (default 36, C1 in Live).
Capture capture	Off / On	Off	A rising edge stores the trace in progress, bypassing the gates and freeze. Same as the Capture button.
Clamp Last (Trigger) trigger_clamp_last	Off / On	Off	Switching it on clamps the newest trace. For MIDI mapping and automation.
Clamp All (Trigger) trigger_clamp_all	Off / On	Off	Switching it on clamps the newest traces up to the clamp budget.
Unclamp All (Trigger) trigger_unclamp_all	Off / On	Off	Switching it on unclamps every trace.
Clear Unclamped (Trigger) trigger_clear_unclamped	Off / On	Off	Switching it on forgets every unclamped trace.
Clear All (Trigger) trigger_clear_all	Off / On	Off	Switching it on clears memory.
Save Memory With Set embed_memory not automatable	Off / On	Off	Save memory inside the Live set (off by default: about 23 MB per minute of stored audio).