Home / Audio / Production and Craft

Affirmology Audio Structures - Formula + Open Work

Updated Jun 26, 2026 · Affirmology_AudioStructures_Formula_and_WIP_v1.md

Affirmology Audio Structures - Formula + Open Work

Operating rule (Jeff, 2026-06-26): every audio structure is documented with (1) THE FORMULA (the locked golden recipe) and (2) WHAT'S STILL BEING TESTED / WORKED THROUGH (open items). "Locked" = golden baseline; it only changes through a deliberate trial → approve → relock cycle. Each structure prevents its own creep.

One structure per section. Add a new section when a new category is defined (The Mirror, sleep/journey, motivational, interludes, etc.).

Structured tracker (the spreadsheet): Affirmology_AudioStructures_Tracker_v1.csv - every parameter as a row with golden_value + status (GOLDEN / IN_PROGRESS / FLAGGED / FLAGGED_PARKED / FIXED) + source date. That CSV is the single source of truth; this doc is the readable companion. Recovered 2026-06-26 by mining all prior sessions so nothing is lost.

Structure: ORIGIN SOUL SONG (the demo / hero identity track)

What it is: each person's own first-person hero-identity track, generated live from their chart. The demo generates a NEW one per visitor - it never plays Jeff's (or anyone's) audio for another user. The asset is the FORMULA, not any mp3.

THE FORMULA (current golden, being tightened)

• Chart-driven, never shared: every sign, gate, gift, line, number, and image comes from THIS person's chart. No other person's content ever appears. (Style is shared; content never is.)
• Voice: Charlotte (ElevenLabs) for the demo (no good replacement yet). Production/scale moves to Fish wherever quality holds, never at the cost of quality.
• Speed: ~0.96 (the proven baseline; NOTE the demo CLI currently hard-codes 1.0 - reconcile).
• Music: Heavenly Circuit bed, ~8s solo intro, voice-anchored mix, QC-gated.
• QC: audio_qc.py runs on every render; a dropout is never served.
• Length target: ~3 - 4 min hero cut (currently the engine default is ~6.3 min / 820 - 950 words - tightening down).
• Structural spine (shadow → gift identity arc): 1. "Listen." opener. 2. Body: lead with the systems/spheres this person's chart is actually about; move each from shadow to gift as present-tense, already-true first-person recognition. Name their specific gates/lines/gifts/signs woven as felt truth, never a roll-call. 3. "I am the one who…" identity declarations from their gifts. 4. Three-noun identity title: "I am the [noun], the [noun], and the [noun]" (their gifts). 5. Embodied landing + a short final beat.
• No em dashes. Poetic priority (image over concept; rooted words over corporate/therapeutic).

STILL BEING TESTED / WORKED THROUGH

• 🔴 Music-fade / sonic ending (PARKED - optimize later, Jeff 2026-06-26): because audios are different lengths, the music outro sometimes lands weak/strange instead of authoritative and strong. The voice-anchored fade needs a consistent, strong close. Not being optimized right now.
• 🟡 Ending text close: 3 candidate signature closes drafted + rendered (Seal / Rise / Landing, Charlotte 0.96) - awaiting Jeff's pick. Whichever wins becomes the formula's consistent close.
• 🟡 Length tighten: bring the default ~6 min down to the ~3 - 4 min hero cut.
• 🟡 Speed reconcile: 0.96 (baseline Jeff recalls) vs 1.0 (demo CLI hard-code) - confirm and lock one.
• 🟡 Spine consistency: the fixed natal-walk spine vs. the flexible Studio path produce different shapes/endings; the formula should distill the best of the proven Jeff origin Soul Songs into one consistent recipe.
• 🟡 Cloud: the demo render must run with Jeff's laptop off (see the cloud North Star + Phase 2 deploy).

VERIFICATION RULE

Prove the formula on two different charts (Jeff + Sol - the active tester; not Colin yet) to confirm it customizes correctly and never bleeds one person's content into another's.

Structure: THE MIRROR (the bonus reading)

What it is: a ~9 - 11 min reading emailed to the demo user the next morning, with their PDF - the second beat of the funnel (Soul Song now → Mirror + PDF next day). Chart-driven, per person.

THE FORMULA (current golden)

• Format: Version A - the "Identity Arc" (SEEN → INSPIRED → LONGING). This is the emotional version that "landed hard." (Version B, "Your 3 Loudest Themes," is the analytical breakdown → it lives in the paid app, not here.) Principle: emotion converts, information retains.
• Voice: MALE - "Dr. Young" / Calm Storyteller (Fish voice id e686ae649ee44f219a108aacba206c1a, s1), slow with sentence pauses. (Soul Song is female; the Mirror is male.)
• Delivery: next morning + the PDF; also a non-buyer re-engagement nudge.

STILL BEING TESTED / WORKED THROUGH

• 🟡 Length: target 9 - 11 min; drafts came in ~6.7 - 7.6 min - needs to stretch to target (add unique depth, not padding).
• 🟢 FIXED: meta-leak at start/end (voiced "…1110 words…" / "render this as a Sacred Audio?") - cleaner hardened to strip word-count/minute preambles and trailing "render this?" lines.
• 🟡 Fish delivery can sound mechanical - needs the slow/paused treatment; revisit voice as Fish quality improves.

Structure: THE PDF REPORT (the Blueprint teaser / lead magnet)

• Status: 🔴 v1 REJECTED - "too salesy/gimmicky, needs real value, craft, care, then teasers."
• Direction: teach-then-persuade arc (StoryBrand customer-hero-journey); sell the Subconscious OS concept + the methodology + the 3 steps; WeasyPrint HTML-to-PDF; no oracle/Sophia/Hermes/council names anywhere demo-facing; on-screen = same data source as the PDF, less busy; leave room for graphics.

Structure: C26 PACING ENGINE (production pacing layer) - GOLDEN / SHIPPED

Deep 0.88, standard 0.95, activation 1.03, ceiling 1.05; real measured silence between segments; bed profiles per structure. Committed 601e7f0, deployed; the demo path is byte-for-byte untouched.

Structure: DAYTIME ANCHOR (~27 - 28 min deep daytime) - IN PROGRESS

Charlotte, --pace deep (0.88), Heavenly Circuit --loop-music, --voice-offset 8.0, QC on. Open: reaching ~27 min needs more unique sections (council only produced ~8 - 11 min of words), not loops/repeats.

Reference (do not confuse with products)

Primer/Trailer VO = ElevenLabs Lily (pFZP5JQG7iQjIQuC4Bku), stability 0.45 / style 0.15 / speed 0.7. Do not regenerate. Not an audio-product voice.

Structure: (future) sleep / journey, motivational, interludes, The Mirror-male-voice variants

Defined as we go. Each gets its own FORMULA + Open Work section + CSV rows.