Generation config with header + grid output (default).

s <- seek "*" $ gen
s <- seek "*" $ cue start $ tonal ctx $ len 4 $ entropy 0.3 $ gen

Generation config with compact musical summary.

Generation config with verbose diagnostic traces.

Static grid: repeats the cue chord for len bars. No database access.

s <- seek "*" $ cue start $ len 4 $ genGrid

genGrid with compact musical summary.

genGrid with verbose diagnostic traces.

Regenerate a range of bars within an existing progression. The cue is inferred from the bar before the start position (wrapping). A cue override applies to every family EXCEPT FStrata: the strata regen must seed from the source's own bar to keep the walk-graph seam valid, so it ignores the override.

FAMILY-AWARE: regeneration produces states of the family the source progression already is — families never mix. The one downgrade is deliberate: hand-built extended (FExtended) sources regenerate as triads (with a printed notice), so the result re-infers as mixed material rather than pretending the fusion family still exists.

• pcProvenance = Just _ — strata-aware path: regenerates all three layers + provenance in lockstep, with one-step lookahead at the e → e+1 seam to keep the spliced bar sequence walk-graph valid under allowedNext.
• FPoly (genE source) — polytonal path: regenerates the foundation range plus both partner chains, seeded from the kept partner bars before the range; the S/M labelling of the source is preserved (chains are never reordered by a partial regen).
• uniform 4-note hand-built material — regenerated bars come back as plain triads with a printed notice.
• uniform 3-note (gen source) — plain triad regen.
• hand-mixed cardinalities — regenerated as plain triads with a printed notice (hand-mixed material is the human aberration channel; regen does not amplify it).

s' <- seek "*" $ entropy 0.3 $ genFrom s 2 3 s' <- seek "*" $ cue start $ genFrom s 2 3 -- override inferred cue s' <- seek "*" $ len 6 $ genFrom s 2 3 -- expand range s' <- seek "*" $ genFrom' s 2 3 -- Standard per-step trace s' <- seek "*" $ genFrom'' s 2 3 -- Verbose trace (+ scoreboard with attempt)

Standard-verbosity alias of genFrom. Mirrors gen', genP' and genI'.

Verbose-verbosity alias of genFrom. Mirrors gen'', genP'' and genI''.

The genE paradigm: polytonal three-layer generation. The T layer is a foundation walk byte-identical to gen (all R constraints apply to it alone — the foundation owns the bass); the S/M layers are partner triad chains, each a corpus-valid walk of its own, sharing exactly 2 pitch classes with the foundation per bar and unioning to exactly 5. The traversal chooses freely per bar between the two admitted geometries — common-dyad (every layer pair sounds 4 tones) and base-anchored (pairs with the foundation sound 4, S+M sounds the pentad). Partners honour key/roots/overtones but never direction specs or strata machinery. Combine layers at the pattern surface via the Layer selectors (TS/TM/SM/TSM/PT). Cues are triadic.

s <- seek "*" $ len 8 $ entropy 0.3 $ genE'

genE with compact musical summary (per-bar layer table).

genE with verbose diagnostic traces (adds partner pool/tier trace).

Generate over the jazz (Change) graph: the genJ family. Same modifier chain as gen (seek "*" $ cue start $ len 8 $ entropy 0.3 $ genJ); the seek spec resolves against BOTH corpora (see Harmonic.Framework.Builder.JazzGen for the full semantics: jazz names blend, classical names steer via the steer dial, "*" walks the whole corpus, "none" is refused — the jazz graph has no offline mode). Progressions carry variable-arity chords (3-6 tones) straight from the corpus vocabulary, stamped FJazz so genFrom regenerates them jazz-natively and attempt ranks them against the Change graph. tonal constraints apply through the same R filter as every family; a step they empty is relaxed with a notice (there is deliberately no fallback pool).

genJ with Standard per-step trace (walk steps, pool sizes, picks).

genJ with Verbose trace (adds seek resolution and steer notices).

Set the genJ classical-steer boost strength (default 3.0). Applies only when the seek spec names classical composers: a jazz candidate containing one of the steer blend's top recommended triads has its score multiplied by up to (1 + strength). 0 disables steering influence entirely. Initial calibration — tune by ear.

s <- seek "debussy" $ steer 6 $ len 8 $ genJ'

Strata-first generation entrypoint. Seeded by a StrataLabel; produces a ProgressionContext with distinct triad, strata, and mode layers and pcProvenance = Just ….

Uncued, the starting chord is drawn from inside the stratum itself — a five-tone set admits few triads, and the ordinary whole-corpus cue almost never lands in one. A cue you pass with cue is honoured as given; one that escapes the stratum is reported with the triads that would have fitted, rather than silently replaced.

s <- seek "none" $ len 6 $ genP VI s <- seek "none" $ cue start $ len 6 $ genP VI

Standard-verbosity variant of genP.

Verbose-verbosity variant of genP.

Silent-verbosity genP aliases, one per Roman numeral — genI pins the starting tristrata to I, genII to II, and so on through genXI.

s <- seek "*" $ attempt 3 12 $ entropy 0.4 $ genI

Three verbosities throughout, by the usual prime convention: genI silent, genI' standard, genI'' verbose.

Silent-verbosity genP aliases, one per Roman numeral — genI pins the starting tristrata to I, genII to II, and so on through genXI.

s <- seek "*" $ attempt 3 12 $ entropy 0.4 $ genI

Three verbosities throughout, by the usual prime convention: genI silent, genI' standard, genI'' verbose.

Silent-verbosity genP aliases, one per Roman numeral — genI pins the starting tristrata to I, genII to II, and so on through genXI.

s <- seek "*" $ attempt 3 12 $ entropy 0.4 $ genI

Three verbosities throughout, by the usual prime convention: genI silent, genI' standard, genI'' verbose.

Silent-verbosity genP aliases, one per Roman numeral — genI pins the starting tristrata to I, genII to II, and so on through genXI.

s <- seek "*" $ attempt 3 12 $ entropy 0.4 $ genI

Three verbosities throughout, by the usual prime convention: genI silent, genI' standard, genI'' verbose.

Silent-verbosity genP aliases, one per Roman numeral — genI pins the starting tristrata to I, genII to II, and so on through genXI.

s <- seek "*" $ attempt 3 12 $ entropy 0.4 $ genI

Three verbosities throughout, by the usual prime convention: genI silent, genI' standard, genI'' verbose.

Silent-verbosity genP aliases, one per Roman numeral — genI pins the starting tristrata to I, genII to II, and so on through genXI.

s <- seek "*" $ attempt 3 12 $ entropy 0.4 $ genI

Three verbosities throughout, by the usual prime convention: genI silent, genI' standard, genI'' verbose.

Silent-verbosity genP aliases, one per Roman numeral — genI pins the starting tristrata to I, genII to II, and so on through genXI.

s <- seek "*" $ attempt 3 12 $ entropy 0.4 $ genI

Three verbosities throughout, by the usual prime convention: genI silent, genI' standard, genI'' verbose.

Silent-verbosity genP aliases, one per Roman numeral — genI pins the starting tristrata to I, genII to II, and so on through genXI.

s <- seek "*" $ attempt 3 12 $ entropy 0.4 $ genI

Three verbosities throughout, by the usual prime convention: genI silent, genI' standard, genI'' verbose.

Silent-verbosity genP aliases, one per Roman numeral — genI pins the starting tristrata to I, genII to II, and so on through genXI.

s <- seek "*" $ attempt 3 12 $ entropy 0.4 $ genI

Three verbosities throughout, by the usual prime convention: genI silent, genI' standard, genI'' verbose.

Silent-verbosity genP aliases, one per Roman numeral — genI pins the starting tristrata to I, genII to II, and so on through genXI.

s <- seek "*" $ attempt 3 12 $ entropy 0.4 $ genI

Three verbosities throughout, by the usual prime convention: genI silent, genI' standard, genI'' verbose.

Silent-verbosity genP aliases, one per Roman numeral — genI pins the starting tristrata to I, genII to II, and so on through genXI.

s <- seek "*" $ attempt 3 12 $ entropy 0.4 $ genI

Three verbosities throughout, by the usual prime convention: genI silent, genI' standard, genI'' verbose.

Standard-verbosity Roman numeral aliases: per-step musical context plus the grid. See genI.

Standard-verbosity Roman numeral aliases: per-step musical context plus the grid. See genI.

Standard-verbosity Roman numeral aliases: per-step musical context plus the grid. See genI.

Standard-verbosity Roman numeral aliases: per-step musical context plus the grid. See genI.

Standard-verbosity Roman numeral aliases: per-step musical context plus the grid. See genI.

Standard-verbosity Roman numeral aliases: per-step musical context plus the grid. See genI.

Standard-verbosity Roman numeral aliases: per-step musical context plus the grid. See genI.

Standard-verbosity Roman numeral aliases: per-step musical context plus the grid. See genI.

Standard-verbosity Roman numeral aliases: per-step musical context plus the grid. See genI.

Standard-verbosity Roman numeral aliases: per-step musical context plus the grid. See genI.

Standard-verbosity Roman numeral aliases: per-step musical context plus the grid. See genI.

Verbose-verbosity Roman numeral aliases: full traces, the grid, and the multi-attempt scoreboard when paired with attempt. See genI.

Verbose-verbosity Roman numeral aliases: full traces, the grid, and the multi-attempt scoreboard when paired with attempt. See genI.

Verbose-verbosity Roman numeral aliases: full traces, the grid, and the multi-attempt scoreboard when paired with attempt. See genI.

Verbose-verbosity Roman numeral aliases: full traces, the grid, and the multi-attempt scoreboard when paired with attempt. See genI.

Verbose-verbosity Roman numeral aliases: full traces, the grid, and the multi-attempt scoreboard when paired with attempt. See genI.

Verbose-verbosity Roman numeral aliases: full traces, the grid, and the multi-attempt scoreboard when paired with attempt. See genI.

Verbose-verbosity Roman numeral aliases: full traces, the grid, and the multi-attempt scoreboard when paired with attempt. See genI.

Verbose-verbosity Roman numeral aliases: full traces, the grid, and the multi-attempt scoreboard when paired with attempt. See genI.

Verbose-verbosity Roman numeral aliases: full traces, the grid, and the multi-attempt scoreboard when paired with attempt. See genI.

Verbose-verbosity Roman numeral aliases: full traces, the grid, and the multi-attempt scoreboard when paired with attempt. See genI.

Verbose-verbosity Roman numeral aliases: full traces, the grid, and the multi-attempt scoreboard when paired with attempt. See genI.

Set starting state.

s <- seek "*" $ cue start $ gen

Set progression length (number of chords).

s <- seek "*" $ len 8 $ gen

Set composer blend and execute. Terminal modifier — produces IO ProgressionContext. Every family returns distinct layers: genP carries curated strata/mode chroma with Just provenance, genE partner triads, and gen/genJ chordscale-derived pentatonic/mode layers (chordscale is applied in the dispatch fallthrough) with pcProvenance Nothing.

s <- seek "*" $ gen s <- seek "bach:70 debussy:30" $ cue start $ len 4 $ gen s <- seek "none" $ cue start $ len 6 $ genVI

Set entropy (>= 0) — the gamma sampler targets rank entropy * 10 in the scored pool, so 0 usually takes the top candidate, 0.5 wanders around the 5th, 1 around the 10th; values above 1 reach deeper still.

s <- seek "*" $ entropy 0.5 $ gen

Set harmonic context (R constraints).

s <- seek "*" $ tonal (hcKey "0#" $ hContext) $ gen

Per-bar position within the dynamically-changing active tristrata. Elements ∈ {1,2,3} cycle circularly. Sets _gcLenOverride to the parsed list length so _gcLen doesn't need to be set explicitly; len applied later clears the override (last-writer-wins).

s <- seek "none" $ relStrata "1 1 2 2 3 3" $ genVI -- 6 bars

Per-bar absolute strata label across all tristratas. Elements are Roman numerals I..XI cycling circularly. Sets _gcLenOverride to the parsed list length.

s <- seek "none" $ absStrata "I V X" $ genI -- 3 bars

Override the same-strata continuity boost multiplier. Values below 1.0 favour candidates whose strata matches the previous bar's. Default 0.90. Pass 1.0 to disable the bias.

s <- seek "none" $ sameBoost 0.5 $ genVI -- strong same-strata pull

Override the flip-flop boost multiplier (candidates matching the grandparent strata when the current /= previous). Default 0.80.

Override the same-tristrata continuity boost multiplier. Values below 1.0 favour candidates whose active tristrata matches the previous bar's. Default 0.70 (strongest of the three).

Run multi-attempt rank-and-select generation: produce up to maxAttempts candidate progressions, stop early once viableTarget viable attempts (all bars ModeOk) have been collected, then return the highest-scoring one. Scoring blends root motion, voice leading, and mode validity via defaultWeightsOffline.

s <- seek "*" $ attempt 3 24 $ entropy 0.4 $ gen -- best of up to 24

Polytonal (genE) attempts rank differently, because they are three progressions rather than one: each layer is scored in its own right (half the weight on the foundation, a quarter on each partner) and a divergence axis rewards the layers standing apart — the point of the family. See PolyScore. Raising maxAttempts raises both the quality and the divergence of what comes back.

Defaults are attempt 1 1 — i.e. the modifier is a no-op when omitted, preserving legacy single-pass behaviour.

Set the viability quality floor used by attempt. An attempt is viable iff psModeValidity >= 1.0 (structural invariant) and totalScore >= floor. Default is 0.5; passing 0.0 recovers the original structural-only viability.

s <- seek "*" $ viability 0.65 $ attempt 3 24 $ gen

Tune downward if attempt N K frequently fails to collect N viable within K (raise K or lower the floor); tune upward if K is being hit consistently with mediocre-quality picks (lower K or raise the floor).

Configuration for the modifier-based generation API.

Built via modifier chains:

s <- seek "*" $ cue start $ tonal ctx $ len 4 $ entropy 0.3 $ gen

Generation mode.

Verbosity level for generation output.

Default generation configuration.

cue:     random root, major triad — but an uncued run resolves its cue
         context-aware at run time: inside the tonal context for gen /
         genE / grid (tonalStartCue),
         from the stratum for genP, from the Change graph for genJ
len:     4
seek:    "*" (all composers)
entropy: 0.2
tonal:   hContext (chromatic)

Execute a GenConfig, producing a progression.

Progression-typed view of execGenConfigPC: one dispatcher and one emission path for every family, and the attempt loop (_gcMaxAttempts) is honoured rather than silently ignored.

Terminal executor producing a ProgressionContext.

When _gcMaxAttempts > 1, dispatches through generateBest for rank-and-select multi-attempt generation; otherwise runs a single pass.

The single-pass dispatch reads _gcMode: StrataMode runs the strata- first traversal producing distinct layers; all other modes fall through to execGenConfig and wrap the resulting triad Progression via fromProgression.

Generate a harmonic progression from a starting state.

Arguments: * start: Initial CadenceState (defines starting root and quality) * len: Number of chords to generate * composerStr: Composer blend string ("bach:70 debussy:30") * entropy: Gamma shape parameter (higher = more unusual choices) * context: HarmonicContext filters (R constraints)

Algorithm: 1. Parse composer weights 2. For each step: query graph, apply R filter, apply E weights, gamma select 3. Apply voice leading optimization to the complete chain

Returns: Progression type (Phase B)

Generate with custom configuration

Simplified algorithm: 1. Start with user-provided CadenceState 2. For each step: build candidate pool, gamma-select next cadence 3. Candidate pool = graph transitions (filtered) + consonanceFallback (unlimited — the pool is never truncated)

Generate a progression returning both result and diagnostics (internal).

This is the core internal function that generates progressions and collects standard-level diagnostics (per-step candidate pools, selections, rendered chords).

For most users, use the unified interface instead: * genSilent - for silent generation * genStandard - for standard diagnostics * genVerbose - for verbose diagnostics

Generate a progression with maximum diagnostic traces (internal).

Like 'generate'' but populates full transform and advance traces for debugging. This function collects: * All standard diagnostics (per-step candidate pools, selections) * Full transform traces (DB intervals, transposition, normalization, zero-form) * Full advance traces (root motion PC arithmetic, enharmonic spelling)

This is SLOWER than 'generate'' due to extra tracing computation. Use only for debugging chord name discrepancies or voice leading issues.

Positional generate with header + grid output (internal).

Positional generate with compact musical summary (internal).

Positional generate with verbose traces (internal).

Generate with custom configuration and maximum diagnostics (internal).

Generate a progression with NO diagnostic output (verbosity 0 - silent mode).

Generate a progression with STANDARD diagnostic output (verbosity 1).

Generate a progression with VERBOSE diagnostic output (verbosity 2).

Print diagnostics collected during generation.

Selects output level based on verbosity parameter:

0 - Silent

No output

1 - Standard

Per-step candidate pools, selections, rendered chords

2 - Verbose

Standard plus transform and advance traces

Useful for manually reprinting diagnostics after extraction from tuple, or batch processing multiple results at different verbosity levels.

Diagnostic information for a single generation step

Complete diagnostics for a generation run

Transform trace captures intermediate values in fromCadenceState → toTriad pipeline. Used for maximum verbosity debugging (gen''). Includes raw DB data plus all transformation stages.

Advance trace captures intermediate values in root advancement. Used for maximum verbosity debugging (gen'').

Harmonic context defines the Rules (R) that constrain the generative space.

These filters are applied BEFORE database evaluation (R in R→E→T pipeline), limiting which cadences can even be considered as candidates.

Three-part filtering system: * overtones: Pitch candidate set (e.g., "E A D G" for bass tuning overtones) * key: Key filter applied to candidates (e.g., C, "#", "bb" for key signature) * roots: Root/bass candidate set (e.g., "E F# G" for valid root notes)

Filters use "*" as wildcard (match all). Format matches legacy Overtone.hs notation.

Constructor for HarmonicContext.

Arguments: * overtones: Pitch set filter ("E A D G", C, "*") * key: Key signature filter (C, "#", "bb", Am, "*") * roots: Root notes filter ("E F# G", "1#", "*")

Example: harmonicContext "*" "*" "*" -- No filtering (all candidates) harmonicContext "E A D G" C "*" -- Bass tuning, C major key harmonicContext "*" "#" "E G" -- G major key, E/G roots only String (not Text) so editor plugins that wrap literals in Tidal's deltaContext (which has a Stringy String instance but no Text one) can pass filter strings straight through.

Default harmonic context for Tidal live coding: all wildcards (chromatic). Named hContext to avoid collision with TidalCycles' EventF.context field.

Use modifier functions to constrain the context:

ctx = invSkip 2
    $ consonant
    $ hcRoots "C E G"
    $ hcKey "0#"
    $ hcOvertones "E A D G"
    $ hContext

Direction of dissonance drift across a generated progression.

When applied to a HarmonicContext, the generation engine filters the candidate pool at each step so that only chords with equal or greater (Dissonant) or equal or lesser (Consonant) dissonance than the current chord are preferred. Free imposes no constraint (default).

Advisory, not hard: if the drift predicate would empty the pool at a step, the filter relaxes and that step proceeds unconstrained rather than reaching an absorbing state (applyDriftFilter in Builder.Core). Note the predicate is dissonanceScore — an evaluation function acting as a filter, the documented E-inside-R leak (see ARCHITECTURE §2).

Set overtone filter. Default: "*" (all pitches).

hcOvertones "E A D G" $ hContext — bass tuning overtones

Set key filter. Default: "*" (chromatic).

hcKey "0#" $ hContext — C major

Set roots/bass filter. Default: "*" (all roots).

hcRoots "C E G" $ hContext — only C, E, G as bass notes

Modify context to trend toward increasing dissonance. Each subsequent chord should have dissonance >= the current chord; advisory — relaxes at any step where it would empty the pool.

Modify context to trend toward decreasing dissonance. Each subsequent chord should have dissonance <= the current chord; advisory — relaxes at any step where it would empty the pool.

Set minimum number of non-inversion states between inversions.

invSkip 0 allows inversions at any step (default, current behaviour). invSkip 1 requires at least 1 non-inversion between inversions. invSkip 2 requires at least 2 non-inversions between inversions. The starting state counts toward the counter (a non-inversion start means the first generated step may already be an inversion with invSkip 1).

Advisory, not hard: if excluding inversions would empty the pool at a step, the spacing constraint relaxes for that step rather than halting generation.

Require specific pitch classes to be present in every generated chord.

Tokens are note names (C, G, Bb). A trailing ? marks a tone as preferred rather than required — it is applied when it does not reduce the candidate pool below a minimum viable size, and relaxed otherwise.

Advisory at the limit: the relaxation chain is preferred → required → unfiltered (applyPedalFilter in Builder.Core), so even required tones are dropped as a last resort at a step where enforcing them would leave no candidates — generation never reaches an absorbing state through a pedal constraint.

hcPedal C $ hContext — C must appear in every chord hcPedal "C G" $ hContext — C and G must both appear hcPedal "C G?" $ hContext — C required, G preferred

Restrict the active tristrata pool for genP.

"" (default) — all 12 tristrata allowed. "5" — lock to a single tristrata (here #5, IV-VI-X). "1 2 5" — whitelist multiple tristrata. "[1,2,5]" — bracket form accepted.

Parsed via parseTristrataList; unknown tokens are silently discarded.

Parse composer string preserving input order Returns list of (name, normalized weight) tuples in order of appearance

Generate portmanteau from composer string (preserving input order) Takes weighted portions from beginning/middle/end based on POSITION Returns Nothing for "*", empty input, or "none" (offline mode)

Extract characters from name based on position in list

Take a leading fragment of a name, sized by the composer's blend weight. Used for the first name in a blend. Always yields at least one character.

Take a trailing fragment, sized by blend weight. Used for the last name in a blend, so the portmanteau ends on a real word ending.

Take a centred fragment, sized by blend weight. Used for names between the first and last in a blend. Ties favour earlier characters.

Print generation header based on composer selection Takes raw Text instead of parsed ComposerWeights, plus entropy