Home / Audio / Production and Craft

Affirmology Sound Design: dynamic, state-attuned beds (plain-language report + action plan)

Updated Jun 21, 2026 · Affirmology_SoundDesign_DynamicBeds_Report_v1.md

Affirmology Sound Design: dynamic, state-attuned beds (plain-language report + action plan)

Why this exists: Jeff said sound is where he feels most ignorant, specifically how to make the sound CHANGE within a track and match different states, instead of one static music bed playing flat under the whole audio. This report demystifies it, gives you the vocabulary to direct professionals, and tells you exactly what to track down now (free or with Suno) versus what to hire out after funding. Sol is generating beds in Suno now, so there are ready prompts at the end.

1. The one idea that unlocks all of it

A "music bed" is not one thing. It is LAYERS played in SECTIONS. - LAYERS (called stems): a finished bed is several tracks summed together, for example a low drone, a warm harmonic pad, an atmosphere/texture, maybe a soft pulse, maybe a small melodic motif. When you have these as SEPARATE files, you can raise or lower each one on its own. That independence is what lets sound breathe and evolve. - SECTIONS (the arc): a Sacred Audio moves through a shape, settle, then deepen, then install the belief, then emerge. Sound should move with it, sparse at the start, fuller in the middle, resolving at the end.

Static today: one bed, one volume, start to finish. The goal: the layers and the volume follow the script's arc. Once you see a bed as "layers I can fade over time," the whole problem becomes manageable.

2. The vocabulary you need (so you can direct an engineer with confidence)

• Stem: one isolated layer of a track (the pad alone, the drone alone, the percussion alone).
• BPM (tempo): beats per minute. Slow (about 50 to 70) for deep and sleep, quicker for activation and gym.
• Key / tonal center: the musical "home note." Beds, entrainment tones, and any melody must share a key or they clash. This is why the C3 tones must be tuned to the bed's key.
• Automation: programmed changes over time (bring this layer up here, fade it out there). This is literally how "changing within a track" is achieved.
• Ducking (sidechain): automatically dipping the music a little whenever the voice speaks, so words always stay clear. Standard for any voiceover.
• LUFS: the loudness standard. We pick a target (about -16 LUFS for spoken-word + ambient) so every audio is consistent and none is too loud or too quiet.
• Headroom: leaving level space so nothing distorts.
• Loop points: where a short piece repeats seamlessly to fill a longer audio.
• Riser / swell / transition: short sounds that carry you from one section to the next.

You now know more than enough to brief a professional. When you say "I want stemmed, sectioned beds per state, tuned to a known key, with loop points and an LUFS target," you sound like you run an audio shop.

3. The two ways to get dynamic sound (and which to pursue)

Option A, COMPOSED SECTIONED BEDS. A composer or Suno writes a piece that already evolves through sections. Easiest to drop in. Least flexible, because the evolution is baked into the file and does not automatically match each person's script length.

Option B, LAYERED STEMS MIXED BY THE ENGINE. We keep a library of stems per state, and audio_mix.py fades them in and out to follow the structure's beats and the actual script length. More setup up front, but it scales to every person and any length, and it lives in the same mix stage as the C3 tone layer, driven by the per-Structure bed_profile from C26.

Recommendation: do BOTH, in stages. Use composed/Suno sectioned beds NOW so we can ship varied audios immediately. Build toward the stem-and-automation engine as the durable system. They are not competing; the stems you gather now feed the engine later.

4. What to track down NOW (free or with Suno, before any funding)

1. SUNO, used well (Sol's current path). Generate INSTRUMENTAL pieces per state, ask for SECTIONS, and if Sol's plan offers stem export, download the stems (Suno separates a track into vocals/drums/bass/other on paid tiers; even partial stems help). Recipes are in section 7. Log keepers in the existing Affirmology_MusicAudition_Log_v1.md.
2. FREE / ROYALTY-FREE LIBRARIES for textures and pads. Look for CC0 or Creative Commons material on places like Pixabay Music, Free Music Archive, ccMixter (music) and Freesound (atmospheres, drones, textures). HARD RULE: check the license on every single file and keep a record (CC0 is safest, CC-BY needs attribution, anything "no derivatives" or "non-commercial" is out for us). When in doubt, do not use it.
3. FREE TOOLS to layer and test by ear. Audacity is free; Reaper is cheap and the industry favorite. You can stack stems, fade them, and export. Our engine already uses FFmpeg, which can mix and fade programmatically, so anything we prove by ear in Audacity we can later automate.
4. SYNTHESIZED ENTRAINMENT TONES, free, generated in-engine. This is the C3 layer (delta/theta/alpha). No purchase needed; it is math, not samples.
5. A SHARED SOUND PALETTE. Pick two or three keys and a small set of instrument characters we like, and keep every bed inside them, so everything we gather mixes together cleanly later.

5. What to hire out LATER (after funding) and the exact brief to hand them

You do not need to become an audio engineer. You need to gather raw parts now and hand a clear brief to two roles later.

ROLE 1, an ambient composer / sound designer. Deliverables to require: - A library of beds per STATE: sleep (delta-friendly), journey/deep (theta), morning (alpha, gently bright), activation/gym (driving). - Delivered as STEMS, not just final mixes (drone, pad, texture, pulse, motif as separate files). - Each piece in a stated KEY and BPM, with clean LOOP POINTS, and provided in a few LENGTHS (or loopable to any length). - Instrumental only, no vocals, no jarring transients; sleep beds with no percussion. - Full commercial license or work-for-hire so we own the raw parts and can reuse them across thousands of renders.

ROLE 2, a mastering / audio engineer. Deliverables: - A master chain and LUFS target for our voice-plus-bed-plus-tone mix, so every audio lands consistent and clean. - Voice ducking settings (how much the bed dips under speech). - A QC standard that extends our existing audio_qc.py gate (we already reject dropouts; they tighten the loudness and clarity bar).

Hand them this report plus the C26 brief and the C3 spec, and they will know exactly what to make.

6. The pieces-to-track checklist (what gets gathered, and by whom)

• Music beds, stemmed and sectioned, per state: Suno now (Sol), composer later.
• Entrainment tone layer (delta/theta/alpha): engine, free, via C3.
• Per-structure pace and bed config: engine, via C26.
• Voice settings (slower for deep types): engine, via C26.
• Transitions / risers between sections: free libraries now, composer later.
• Ducking + mastering + LUFS target: rough by ear now, mastering engineer later.
• License records for every gathered file: ongoing, non-negotiable.

7. Suno prompt recipes for Sol (use now)

Generate instrumental, no vocals. Specify mood, instrumentation, tempo, evolution, and length. Ask for sections so the piece evolves. Export stems if the plan allows.

SLEEP / before-bed: "Ambient instrumental for deep sleep and hypnosis, no vocals, no drums, no percussion. Slow 55 BPM, key of C. Warm low drone and soft analog pads, distant and spacious, very gentle. Begins sparse, swells slightly in the middle, dissolves to near silence at the end. Calm, safe, womb-like. 3 minutes, seamless loop."

JOURNEY / deep meditative: "Cinematic ambient meditation instrumental, no vocals. 60 BPM, key of D minor. Theta-state feel, deep evolving pads, subtle low pulse, faint shimmering texture that grows then recedes. Three sections: settle, deepen, return. Mysterious but safe. 4 minutes."

MORNING / calm and aware: "Uplifting but calm ambient instrumental for a morning meditation, no vocals. 70 BPM, key of A. Warm felt-piano motif over soft pads, gentle brightness, hopeful and grounded, builds slightly toward the end. Clean, modern, not cheesy. 3 minutes."

ACTIVATION / gym / motivational: "Driving cinematic instrumental for motivation, no vocals. 90 to 100 BPM, key of E minor. Pulsing low synth, steady rhythmic energy, rising momentum across the track, confident and powerful, not aggressive. Builds to a strong but controlled peak. 3 minutes."

Tip for Sol: generate several per state, keep the ones with the cleanest low end and no sudden loud moments, and log them. Variety of KEYS is fine as long as each individual track is internally consistent; note the key in the log so we can match tones later.

8. Your "free courses / information" question (yes, and a lot of it)

Two kinds of free material, treated differently.

A. PUBLIC-DOMAIN TEACHERS we can ingest FULLY into the corpus (old enough to be free of copyright). These are gold and free: Neville Goddard (most lectures), Wallace Wattles, Florence Scovel Shinn, Thomas Troward, Genevieve Behrend, James Allen, Emile Coue, the Kybalion, and the older hypnosis literature. Public domain means we can fold the actual text in, not just benchmark it. This directly answers your question and it is a big, free win.

B. FREE LEARNING for you and me (study, not ingest): university open courseware, free lecture series, and reputable free explainers on the neuroscience of state, hypnosis induction craft, and audio production (mixing voiceover with music, sidechain ducking, LUFS for spoken word, ambient composition). We study these to sharpen the craft; we do not copy in-copyright material into the corpus.

THE RULE that keeps us safe: public-domain works get ingested; in-copyright authors (and branded programs like RTT) are studied as benchmarks and distilled into our OWN original method, never reproduced. This is the same tier-wall we already apply.

I am folding the public-domain list and the "research free courses" instruction into the C28 research lane, so the nightly builder prioritizes the free, ingestible sources first.

9. The honest bottom line

You are not behind because you lack audio skill. You are exactly where you should be: gathering raw parts (Suno beds, free textures, public-domain wisdom) and writing clear briefs, so that when funding lands you hand a composer and a mastering engineer a precise spec and get back a reusable library. The engine work (C26 pacing, C3 tones, the stem-mixing future) is mine and Code's to build. Your job is taste and direction, and the prompts and briefs here let you do exactly that.