Home / Audio / Production and Craft

Affirmology Sound Layering: The Plain-English Playbook (LIVING)

Updated Jun 25, 2026 · Affirmology_SoundLayering_PlainEnglish_Playbook_v1.md

Affirmology Sound Layering: The Plain-English Playbook (LIVING)

Status: living, started 2026-06-21 (Cowork). This is Jeff's home base for HOW we physically build the audio: the bricks in plain words, the cheap duct-tape way to do it now, the biggest-bang-for-the-buck wins, a parking lot to drop outside info and links, and a go-get task list. No sound-engineer background needed. Written at a deliberately simple level on purpose. We add to it every time we learn or buy something new.

Companion to Affirmology_AudioSuites_BuildQueue_v1.md (WHAT audios we are building) and Affirmology_LayeredStemEngine_OptionB_Spec_v1.md (the engine that assembles beds). This doc is the HOW, in human language.

The one picture to hold: a singer and a band

Every audio is a SINGER with a BAND behind them. - The SINGER is the VOICE: the affirmations, the words, the recognition. - The BAND is the BED: everything playing underneath. The entire craft is making the band support the singer without ever stepping on the words.

The most important thing in the whole audio is FREE: the identification. The name, the real identifiers, the chart truths said back to a person. That is what makes a tester say "yeah, I get it." The bed just has to carry the words, not show off. Expensive sound is polish on top of a thing that already works because of the words.

The bricks, in plain words

• VOICE. The spoken affirmations. The star of the show.
• BED. All the music and sound under the voice.
• LAYER (also called a STEM). One ingredient of the bed playing alone. A low hum, a soft chord wash, a bit of shimmer, a gentle pulse. We make these one at a time so they stay clean.
• DRONE = a steady low hum. The floor of the bed.
• PAD = soft swelling chords. The warmth.
• TEXTURE = airy shimmer, distant atmosphere. The space.
• PULSE = a gentle steady beat. Only for the energizing ones (gym, walking).
• MOTIF = a few sparse piano-ish notes with space between them. Optional sparkle.
• LOOP. A short clip that repeats with no audible seam, so twenty good seconds can fill twenty minutes. Like a looping background GIF for sound.
• DUCKING. The software automatically turns the bed down a touch while the voice speaks, and back up in the gaps, so the words always sit on top. Our engine already does this. Free pro-sound trick.
• BED ARC. The bed getting sparser or fuller across the audio. Empty and calm at the open, fuller in the middle where the real work happens, fading to near silence at the end of a sleep one. Just volume moving on a timeline.
• CROSSFADE / SEQUENCING. Playing one audio into the next as a single journey (breath opener into install into sleep close), blended so one melts into the next. This is how a "track" is more than one file.
• TONES. Binaural beats and solfeggio / sacred frequencies. A low hum or felt pulse added into the bed for feeling. We add them for the experience and stay honest the science is soft. Off by default.
• BPM (beats per minute). How fast the pulse is. Slow, around 55 to 60, for calm and sleep. Brisk for the gym.
• KEY. The musical "home note" (like "key of C"). If every layer is in the same key, they stack without clashing, and the tone can be tuned to match.
• LOUDNESS (LUFS). How loud a clip feels. We even these out so one layer does not jump out over another.
• BEAT-ANCHORING. Landing a key affirmation right on a drum hit or a singing-bowl strike so it lands harder and gets remembered. Advanced. Save for after funding.

How the pieces snap together (the assembly line)

1. MAKE THE ELEMENTS. Get clean single layers out of Suno, one at a time (drone, pad, texture, maybe pulse), all in the same key. Prompts live in the Option B spec and the Suno prompts doc.
2. LOOP THEM. In Audacity (free), trim each to a clean stretch, set a seamless loop, fade the very ends. Now each layer can run any length.
3. STACK INTO A BED. Layer the drone, pad, and texture together. For now we can do this by hand once per state to get one finished bed. Later the engine does it automatically and makes it evolve.
4. THE ENGINE DUCKS AND ARCS. The voice goes on top, the engine ducks the bed under the words and moves the arc across the sections, and adds a tone if we want one.
5. STITCH INTO A TRACK. If we want one audio to flow into the next, we crossfade the files into one journey.
6. QC. The quality check rejects anything with a dropout or clipping before it ever reaches a listener.

NOW vs LATER (the honest two-tier plan)

NOW (funds low, duct-tape, still hits "yeah I get it"): - Suno (already paid, $0 extra) for clean single-element ambient loops. - Audacity (free) to trim, loop, and fade. - The engine's existing ducking (free, already proven on the demo). - One simple hand-mixed bed per state (sleep, journey, morning, gym) to unblock production this week. - The right BPM and a warm clean loop. That is enough under a strong voice. - A simple drone or solfeggio tone under the sleep ones (cheap to generate), with the honesty note. - Lean hardest on the FREE superpower: the name, the identifiers, the recognition.

LATER (as money comes in, in rough order of payoff): - Commissioned owned signature beds per state (the Affirmology house sound; Sonic Spheres quote already parked, deferred until funded). - Real binaural / isochronic / sacred-tone design done properly. - Timing affirmations onto beat points, singing bowls on the sleep ones, hypnotic music married to the words. - Professional mastering for loudness and warmth. - The Option B engine assembling evolving multi-layer beds automatically (depends on C26, then C3, then C30).

Smallest effort, biggest bang (RANKED, living)

The cheap wins we do along the way. Reorder as we learn. 1. SAY THE NAME + STRONG IDENTIFIERS. Free. The single biggest hit to "this was made for me." Already proven on Jeff this week. 2. ONE CLEAN HAND-MIXED BED PER STATE. A few hours in Suno + Audacity. Unblocks the whole suite. 3. RIGHT BPM PER TYPE. Free, just a prompt and a tempo choice. Slow for sleep, brisk for gym. 4. ENGINE DUCKING. Free, already built. Makes amateur layers sound intentional. 5. A WARM LOW DRONE UNDER THE VOICE. Cheap. Adds "container" feeling for almost no effort. 6. SILENCE AT THE CLOSE OF SLEEP AUDIOS. Free. Lets the install travel into sleep. 7. SIX-BREATHS-PER-MINUTE OPENER. Free craft. Settles the body so the words land. 8. A SINGLE SOLFEGGIO / TONE LAYER ON THE SLEEP ONES. Cheap to generate. Feeling, with the honesty note. (Add new cheap wins here as we find them.)

Learning the sound language (reference-track analysis tools + workflow)

The problem: hard to get what we want out of Suno because we do not yet have the words. The fix: tools that study a track we like and hand back descriptive language, which we translate into better prompts. (Analysis only, benchmark not clone, the same study-never-reproduce rule we use everywhere.)

TWO KINDS OF TOOL (researched 2026-06-21): - REVERSE-PROMPT (clip in, Suno-style prompt out): JustBuildThings AI Music Prompt Generator, Suno Meta Tag Creator, OpenMusicPrompt, SunoPrompt. Mostly free / freemium. Fastest "I like this" to "here is language to ask for it." - ANALYSIS / TAGGING (rich description so we build vocabulary): Sonoteller.ai (free, paste a YouTube link or upload, returns genre/subgenre/mood/instruments/BPM/key), Cyanite.ai (pro subscription, tags + natural-language descriptions + similarity search, the serious B2B one), BeMusic AI (free BPM/key/genre). - TECHNICAL FINGERPRINT + STEMS: Moises (~$4 to $12/mo): BPM, key, chords, and stem separation (also a paid alternative to Demucs for pulling layers out of a finished rhythmic track, ties to Option B path 2).

THE WORKFLOW (tuned to beds, not full songs): 1. Take a reference clip you love. 2. Run it through a reverse-prompt tool for a draft description. 3. Confirm BPM and key with Moises or Sonoteller. 4. Read Sonoteller's instrument and mood tags for richer words. 5. Translate into our ISOLATED-ELEMENT grammar, one layer at a time (a warm pad in that key, a low drone, no drums), not the whole track. That is how the language becomes better Suno beds for us.

TO BUILD NEXT: a sound-description glossary in this playbook (instruments, textures, production terms, mood words, BPM ranges for our bed states) so we can describe what we want without a tool over time.

Reference beds (Sol's picks) + target specs - analyzed 2026-06-21

Three beds Sol likes, fingerprinted in the sandbox (numpy + ffmpeg). Key estimates solid; tempo on ambient is approximate. Add Sol's + Jeff's FEEL words per track to complete the picture (the machine gives specs, not feel). - JOURNEY (9:00): F major, slow ~62 BPM, fairly bright with shimmer up top, EVOLVING (does not loop, must be extended). A moving, airy journey piece. - MORNING AWARE (10:24): A minor, ~86 BPM, warm and low-mid heavy, little top end, BUILDS across its length. Note: minor and emotive, not bright-cheery. A real taste signal. - SLEEP DEEP DRIFT (8:00): C major, very steady/even, warm, the MOST LOOPABLE (near-flat, loops cleanly). Textbook bed shape. - TASTE THROUGH-LINE: warm, slow, low-mid rich, not bright or synthy. Use as the house-sound target. - STARTING PALETTE (grounded in the above): sleep = C, steady; journey = F, slow + airy + evolving; morning = A minor, warm, building ~86.

Extendable / loopable beds from Suno (for long journey, daytime, Gene Keys audios)

Long audios (9 to 25 min) need beds that extend. Three ways: - BEST on Pro: do NOT ask Suno for one long finished track. Ask for a short STEADY loopable ambient section (drone + pad, no drums, no arc) in the target key, keep the cleanest 1 to 2 min, then loop/extend to length in Audacity. Steady material (like Sol's Sleep bed) loops seamlessly; evolving material (like her Journey bed) does not, it must be Extended. - Suno EXTEND feature: lengthens a piece directly. Simple, but the feel can drift across extensions. - Premier (Suno Studio) TEMPO-LOCKED WAV: the cleanest loop export, if/when we upgrade. - The engine (Option B) loops/stretches a layer to any script length AND adds the rise-and-fall arc, so a looping base never feels monotonous over a 20-min journey. - MOVE: use Sol's three as the target spec, generate steady loopable elements in those keys, let looping + the engine cover the length.

Bed-production tools (owned instruments vs AI generation)

Two complementary ways to make beds. Use both. - TRIBE INSTRUMENTS "Sound Healing" VST (~$124.99 one-time, Jeff surfaced 2026-06-21): 19 sound-healing instruments + 330 presets, runs in a DAW (GarageBand, free on Mac, is supported, so no pricey software). Palette maps straight to our roles: gong bowed-sustain / didgeridoo / Healing Synth / Feedback drones = DRONE + PAD; ambiences / wind chimes / rainsticks / noise = TEXTURE; bowls / handpan / kalimba = accents (and the bowls-on-beat idea). STRENGTH over Suno: control. Hold an exact note in an exact key as long as you want, loop it, tune it, own the output. Best tool for DEEP / AMBIENT / sound-healing beds (sleep, journey). CATCH: it is hands-on, not API/AI-automated. Someone plays a preset in GarageBand, records a loop, exports it. Those exports become high-quality LAYERS in our stem library (feeds Option B). Confirm the EULA allows commercial-product use before leaning on it (standard for sample libraries). - SUNO: keep for quick variety and anything RHYTHMIC, and for fast experimentation. AI-prompted, less control, but fast. - THE SPLIT: Tribe VST = controllable, authentic, owned beds (the craft path). Suno = fast, varied, prompted beds. Both export loops into the same stem library.

Outside-info parking lot (Jeff drops it here, Cowork sorts it)

Paste any link, track, tool, course, or find here with a word or two of what it is. Cowork files it into the right slot above and tells you if it is a now-win or a later-buy. - (empty: drop links and finds here)

Go-get tasks (concrete errands that make an audio better)

The "go get this one piece" list. Each is a small, specific fetch that adds a brick. Cowork keeps this current as we design. - [ ] Sol: generate clean single-element ambient loops in Suno, one per role (drone, pad, texture) in a locked key, for the SLEEP and MORNING states first (the two we need first for the wealth suite). Keep the cleanest of several tries. - [ ] Jeff: send your heart-coherence breathwork so we can render the opener and slot it into the morning and nightly audios. - [ ] Sol: install Audacity (free) so we can trim and loop the Suno elements. - [ ] Decide one KEY and one BPM per state and write them down, so every layer matches (start: sleep = key of C, ~55 BPM; morning = brighter key, livelier feel). - SUNO PLAN: start on PRO ($10/mo, 2,500 credits ~500 songs). It already gives the two things this task needs: commercial rights + stem separation with WAV download (Auto Split 12 stems, Split from Mix). Upgrade to PREMIER ($30/mo, 10,000 credits) only when (a) producing past ~500 songs/mo, (b) clean looping in Audacity gets fiddly and Premier's TEMPO-LOCKED WAV would save real time, or (c) we want Suno Studio (multitrack DAW, MIDI, Advanced Split = pull any of ~100 instruments). Premier is volume + depth, not a task unlock. - (Add as we go.)

Open questions / to-learn (kept open on purpose)

• How clean can Suno really get a bare drone vs how much Audacity cleanup it needs. Test and note.
• Whether one hand-mixed bed per state is enough for the beta, or if testers will want variety. Find out from the first listens.
• Loudness target so the voice always wins. Set a number once and reuse.
• (Add as questions come up. The hypnotherapy research landing tonight may add craft about how music and pacing anchor a suggestion; fold it in here.)