Drone Meditations

1Quick start

Tap Play. Tweak. Breathe.

1. Open the app (web at dronemeditations.com or the iOS app).
2. Press the round play button at the bottom. A 30-min session timer starts; audio fades in.
3. Try the PRESET pill — start with "Earth Drone" or "Tibetan Bowl."
4. To explore: drag any oscillator's frequency slider; toggle S (solo) and M (mute) to listen to one voice at a time.
5. Tap the screen anywhere outside controls to hide the panel and watch the Chladni visualization respond.

2Anatomy of the screen

Four floating zones: header, pill row, oscillator strips, transport.

From top to bottom: header (app name, tuning + chord summary, spectrum / camera / cymatics-style toggles) → pill row (CHORD · PRESET · DRIFT · LISTEN · PERFORM · JOURNEY · MORPH · GALLERY) → oscillator strips (one per voice, each carrying waveform + frequency + pan + level + filter + FX + 5 LFOs) → transport (Play / Stop / Record + session timer) → copyright + Manual link (the slim row at the very bottom).

Tap or click anywhere outside a control to dismiss the panel and reveal the visualization full-screen. Tap again to bring it back. A slim "Tap to show controls" hint appears at the bottom when hidden, with mini freq + S/M controls so you can still adjust audio while watching the Chladni.

3Oscillator strips

Four independent voices, each routed through its own FX chain to the master limiter.

Per-strip controls

Filter

Each voice has a state-variable filter with selectable mode (LP, HP, BP), cutoff (20 Hz – 8 kHz, log), and Q 0.3 – 20.

Q = resonance. They're the same control under two names — Q is the technical term (quality factor: cutoff ÷ bandwidth), resonance is the common synth term. Higher Q narrows the filter peak around the cutoff, which we hear as a taller, more whistling/squelchy emphasis at that frequency. Low Q (0.3–1) = smooth pad shaping; high Q (5–20) = vowel-like resonant sweeps.

4Chord, key, octave

Press the CHORD pill to generate all 4 voice frequencies from a chord template.

Choose a root note (C, C♯, D…), an octave (1–6), and a chord type from one of these categories:

• Common — Major, Minor, Sus2, Sus4, dim, aug, Maj7, m7, 7, 9, …
• Extended — 11, 13, add6, m9, maj9, mMaj7, 7♭9, 7♯11, …
• Modal — characteristic 4-note slices of each diatonic / harmonic / melodic mode: Ionian, Dorian, Phrygian, Lydian, Mixolydian, Aeolian, Locrian, Harmonic Minor, Melodic Minor. Each entry captures the mode's most identifying degrees (e.g. Phrygian = 1 / ♭2 / ♭3 / ♭7). When Quantize to scale is on, the snap cache fills with those notes so a pitch-LFO arpeggiates inside the mode instead of wandering chromatically.
• Cymatic — Tetrachord 4:5:6:7, 3:4:5:7, harmonic 6:8:10:13, etc. Chosen to give visually striking nodal patterns.
• Spectral — Sable's chord (Vingt Regards), Scriabin Mystic, Ligeti Lontano, …

5Tuning systems

The frequencies the chord pill produces depend on the tuning you choose.

• 12-TET — modern equal temperament (default). 12 equal semitones per octave.
• Just Intonation — small-integer ratios. Beats less; sounds "locked."
• Pythagorean — stacks of 3:2 fifths. Bright, edgy.
• Verdi (432 Hz) — A4 = 432 Hz instead of 440.
• Lou Harrison Free JI — flexible just intervals tuned to a 1/1.
• Harry Partch 43-tone — Partch's 11-limit microtonal lattice.
• Wendy Carlos α (alpha) — 78¢ step, ~15.4 notes per octave.
• Wendy Carlos β (beta) — 63.8¢ step, ~18.8 notes per octave.
• Wendy Carlos γ (gamma) — 35.1¢ step, ~34.2 notes per octave.

Tunings only affect how chord frequencies are computed; the per-voice frequency sliders remain free continuous controls so you can hand-tune outside the lattice anytime.

6Presets & voice presets

Two kinds: full presets (all 4 voices + master + FX) and per-voice voice presets (one strip at a time).

Full presets

Press the PRESET pill. Categories include:

• INIT (pinned to the top) — a neutral starting patch. Four sine voices at A2 / E3 / A3 / C♯4, modest pan spread, all FX zeroed, all LFOs off. Identical to the state the app boots into on first launch. Use it as a blank canvas when you want to build a sound from nothing instead of editing a character preset.
• Drone Artists — 38 character-rich presets that capture each artist's signature waveforms + FX + LFOs, not just their pitches: Pauline Oliveros, Terry Riley, Éliane Radigue, Stars of the Lid, Sunn O))) (drop master volume first), William Basinski, Phill Niblock, Charlemagne Palestine, Yoshi Wada, Harold Budd, Alice Coltrane, Earth, Nurse With Wound, Keiji Haino — most appear twice, once as the original tribute and again as a (Granular) or (Fading) variant that exploits the new granular waveform and per-loop sample fades. A final Drift Showcase sub-set (Tide Breath, Detune Choir, Quarter-Tone Cluster, Pendulum Dawn, Tibetan Bowl Cycle, Microtonal Dance) demos the per-voice drift amount / period overrides — each preset uses a different drift shape so you can hear what the Ocean / Wave / configurable-period modes actually do.
• Developer Patches — 16 patches showcasing the granular sampler, BPM-quantized grain density, FX Mix LFO, modal chords, and per-voice timing envelopes. Includes four named JG patches converted from a personal collection (Ondulations, Dub Wave, Freedom Did Not Lose, Piano Repetition), 10 procedural patches in the same vein (built around unusual scales, modal tunings, granular sampling, and staggered start delays), and two showcase patches (Pulse Lock, Bloom Swell) that lean specifically on BPM grain sync + FX Mix LFO swells. Treat this section as the demo-reel for "what the new tools can do."
• Binaural — Theta, Alpha, Beta and Gamma beat presets (paired carriers ~7, 10, 17, 40 Hz apart).
• Natural Resonance — Earth, OM 136.1, Sun 126.22, Moon 210.42, Phi-augmented chord, Sable's chord.
• Cymatics — chords chosen to produce especially symmetric Chladni patterns.
• Mystic & Composers — Scriabin Mystic chord, Ligeti chromatic / whole-tone / microtone clusters.
• Solfeggio — 96, 174, 285, 396, 417, 528, 639, 741, 852, 963 Hz singles. See the note on Solfeggio claims below.

Tap "Save as preset…" to capture the current full state under a custom name. User presets live alongside the factory list in their own section.

Voice presets

The ⭐ button on each oscillator strip opens a menu to Save current, Load…, or Delete… a single-voice snapshot — waveform, freq, level, pan, filter, all 5 LFOs, FX, drift. Useful for keeping a perfect bass and quickly swapping the pad above.

Sharing presets between devices — iPhone ↔ iPad ↔ browser ↔ a friend

Every saved user preset can be packed into a .dronepreset file — a single JSON envelope containing the full per-voice state plus any loaded sample audio inline. The receiving device reconstructs everything end-to-end, no extra copy of the audio needed.

• Share (iOS): in the Presets sheet, tap the ↑ icon next to a preset. The iOS share sheet opens — pick AirDrop, Save to Files, iCloud Drive, Mail, or Messages. Typical preset is ~5–15 KB without a sample, ~7 MB with a 5 MB WAV.
• Import (iOS): top-right ↓ icon in the Presets sheet opens the Files picker filtered to .dronepreset. Or — even simpler — just tap the file in Files / Mail / AirDrop and iOS hands it directly to Drone Meditations.
• Share (web): the ⇪ button on each row downloads the file. Drag it to a Files window or iCloud Drive folder and it appears on every device signed in.
• Import (web): top-right Import… button in the Presets sheet opens a file picker; pick a .dronepreset and it's added to Your Presets.
• Same format on iOS and web — a preset shared from the iPhone app loads in the browser app and vice versa.
• Re-imports never overwrite — every import gets a fresh id, so importing the same file twice gives you two distinct entries. The original creation date is preserved so you can see when the author saved it.

For automatic iPhone ↔ iPad sync without manual sharing, see the iCloud preset sync note in the iOS app's Presets sheet (presets you save on one device appear on the other within a few seconds; samples still travel via .dronepreset files).

7Waveforms

Eight source options per voice — four periodic, three stochastic (incl. granular), plus user samples.

Noise voices ignore the frequency slider and just hiss. They still pass through every per-voice control downstream: filter (HP for airy hiss, BP for vowel-noise), drive (warmer texture), chorus, delay, reverb, LFOs (S&H on cutoff gives the gritty pattern shifts in the Nurse With Wound preset). Pink noise is generated by Paul Kellet's economy variant of the Voss-McCartney filter; white is uniform random.

Granular synthesis (the Granular waveform)

When you pick Granular as the waveform, a GRAIN row appears under the picker with four sliders:

• SIZE — grain length, 5–500 ms (log). Short grains read as clicks; medium grains (~25–80 ms) sound like raindrops; long grains (~200–500 ms) bleed into soft clouds.
• DENSITY — average grains per second, 0.5–50/s (log). Below 1/s the readout flips to N/min for the geiger-counter range. Combine sparse density + short grains for "rare events," or dense + medium for a downpour. BPM-quantized density: tap the small denomination chip next to the slider (it cycles off → ½ → ¼ → ¼t → ⅛ → ⅛t → 1/16 → 1/16t → 1/32 → 1/32t) to lock grain spacing to a musical division of the global BPM. The slider then becomes a visual indicator only — the actual density follows tempo. Change the BPM and every quantized voice retimes instantly. Combine 1/16t grains on one voice with ¼ grains on another for polyrhythmic clouds that still feel locked.
• JITTER — randomness in inter-grain timing, 0–100 %. 0 = perfectly clocked; 100 = pure Poisson scatter (each grain feels independent).
• SPREAD — random per-grain stereo placement, 0–100 %. Acts on top of the voice's main pan slider, so you can offset the grain field while keeping its center.

The source material is pink noise (warmer than white for meditation). Filter and reverb apply downstream, so a high-pass + plate-reverb stack turns granular noise into raindrops landing in a glass-walled room. Stack two granular voices panned opposite for stereo rain showers — the Sparse Rain and Rain Shower presets do exactly this.

8Per-voice FX — filter, drive, FM, chorus, delay, reverb

Each oscillator has its own filter + drive + FX chain — vertical order on the strip is FM → Chorus → Delay → Reverb. Drive lives next to the filter since it shapes the source before the rest of the chain.

The feedback slider controls how many echoes you get. Above ~0.85 it starts feeding into itself — use sparingly.

9The 5 LFOs per voice

Each voice has five low-frequency oscillators. LFOs 1–4 share a common pool of synth targets; LFO 5 has its own dedicated row of granular + delay + reverb targets — so a single patch carries up to 20 multi-target modulators.

Shared controls (all 5 LFOs)

• Shape — Sine, Triangle, Sawtooth, Ramp, Square, Sample & Hold.
• Rate — 0.02 – 8 Hz (log). Each LFO can also lock its rate to a musical subdivision of the global BPM (½ → 1/32T).
• Depth — how strongly it modulates the target.
• Target(s) — toggleable chips: one LFO can drive multiple targets at once. Tap PAN and CUT on the same LFO and a single triangle wave will sweep both stereo position and cutoff in lock-step.

LFOs 1–4 — synth targets

Each of the first four LFOs picks from a shared seven-target pool:

• VOL (amplitude), PAN, CUT (filter cutoff), Q (filter resonance), Pitch, FM (FM index), and FX (FX Mix macro).
• The FX target is a macro: it biases the voice's reverb, delay, and chorus mix amounts together (±0.5 around their base values), so a slow sine on FX produces a single coordinated wet/dry breathing swell across the whole effects rack — much more cohesive than independently modulating each wet-mix slider. Pair with a fast pitch-LFO or grain texture for dub-style "let it ring out, pull it back" gestures.

LFO 5 — granular + FX targets

LFO 5 sits in its own row right below LFO 4, with a two-row chip grid grouping its nine dedicated targets. None of LFOs 1–4 can reach these — they're reserved for LFO 5 so the lower LFOs stay focused on synth modulation and the rhythmic FX gestures live on their own line.

^† iOS only. On the web app, dFB, dMix, and rDec chips are present and store correctly in .dronepreset files, but the underlying Web Audio nodes can't be modulated cleanly per-buffer the way the iOS DSP can, so they no-op at the audio layer. iOS presets sharing these targets remain compatible — they just sound a little simpler on web.

Why LFO 5 exists. Bumps the modulator count from 16 to 20 per patch, but the bigger reason is texture: stuttering grain pulses, grain-rate accelerandos, two grain clouds at slightly different rates beating against each other, delay-time chorusing, breathing reverb decay — all things you couldn't do with LFOs 1–4. Try a square shape on LFO 5 at gDens with the density chip locked to 1/8 — instant rhythmic grain pulse without touching any sliders.

10Samples — upload + bundled

Any of the four voices can play back a looped audio sample instead of a generated wave. Two ways to load one:

Bundled samples

The Bundled ▾ button on each oscillator strip opens a small picker of audio files shipped with the app. Tap one and it loads instantly — no upload dialog, no Files-app round-trip. Web sources files from web/samples/ via a manifest (web/samples/index.json); iOS sources from a Samples folder inside the app bundle. Both folders have a README explaining how to drop in your own audio. The picker groups entries by an optional category (e.g. "Acoustic", "Cosmic", "Field", "Instruments").

Anything you capture with the per-voice 🎙 Record sample button (see §3) automatically appears at the top of the picker under a Recorded Samples group, so you can drop your own recordings into other voices without re-recording. Recorded samples persist across launches.

Adding to the bundled list — for maintainers + contributors

iOS and web are kept in sync by a small Python script. New samples land on iOS first (where they auto-include in the next Xcode build), then a one-line command mirrors them to web and regenerates the picker manifest.

1. Drop the file in iOS first. Put it in DroneMeditations/Samples/<Category>/. Pick an existing category folder (Acoustic, Atmospheric, Cosmic, Developer, Field, Instruments, Urban) or make a new one. Supported formats: WAV, MP3, M4A, AAC, OGG, FLAC, AIF/AIFF. Filename without the extension becomes the default display name.
2. Run the mirror script.

   python3 web/samples/mirror_from_ios.py

   It copies every audio file from DroneMeditations/Samples/ into web/samples/<Category>/, deletes web entries that no longer exist on iOS, and regenerates web/samples/index.json fresh. Pass --exclude-larger-than 10 to skip files over 10 MB (or whatever cap you want) if a particular sample would inflate the static site too much.

3. (Optional) pretty up the display name. The script auto-titles obvious kebab-case stems (phi-drone → Phi Drone) and passes through already-pretty names (Bansuri B2). For everything else — em-dashes, ampersands, branded casing — add a row to web/samples/display_names.json:

   {
     "overrides": {
       "your-file-stem": "Your Pretty Display Name"
     }
   }

   Re-run the script after editing.

4. Commit + push. The script makes the iOS folder, the web folder, the manifest, and (optionally) the display names file all change together — one commit covers it.

Converting a .dronepreset into a built-in preset — for maintainers + contributors

Saving a preset on iOS or web gives you a .dronepreset file you can AirDrop or import elsewhere. Promoting one of those personal saves into a built-in preset (the kind that ships in the app's Developer Patches / Drone Artists / etc. categories) needs a two-script pipeline that extracts the embedded audio and generates the matching Swift code. Just dropping a .dronepreset into a folder does not auto-pick-up on the next launch.

1. Drop the .dronepreset into the staging folder. Put the file in JGPresets/ (named for the original "JG…" patches but used as the staging area for any incoming preset envelope, regardless of who authored it).
2. Extract the embedded sample(s).

   python3 JGPresets/extract.py

   Walks every .dronepreset in the staging folder, pulls the inline-encoded WAV bytes out of each envelope, and writes them into DroneMeditations/Samples/Developer/<canonical-name>.wav. Samples are content-hash-deduped — if two presets share a sample, the second one is matched against the first by SHA-256 and not re-written. The script also refreshes JGPresets/_summary.json with the preset metadata + which canonical sample names each voice references. Pass --max-mb 60 (default) to skip envelopes whose embedded audio is larger than the cap.

3. Tell the Swift generator to include the new preset. Edit JGPresets/gen_swift.py and add the preset's name to the INCLUDE set, plus a one-line subtitle to SUBTITLES:

   INCLUDE = {"JG Dub Wave", "JG Ondulations", "JG JoG", ...}
   
   SUBTITLES = {
     ...
     "JG JoG": "Author recording · JoG sample-granular study",
   }

   Names that aren't in INCLUDE are silently ignored — useful for staging presets you haven't curated subtitles for yet, or that have huge samples you want to trim before bundling.

4. Generate the Swift block.

   python3 JGPresets/gen_swift.py

   Writes JGPresets/_generated_presets.swift — a paste-ready list of Swift Preset(…) initializers, one per included preset. Each block translates the JSON envelope into typed Swift fields (Voice, FilterState, ReverbState, etc.) so it reads naturally inside Preset.swift.

5. Paste the block into Preset.swift. Open DroneMeditations/Models/Preset.swift, scroll to the Developer Patches section (look for the comment "Generated from JG .dronepreset files…"), and paste the newly-generated block in alphabetical order. The Xcode Samples/ folder is a folder reference, so the new WAV ships in the app bundle on the next build with no project.pbxproj edit needed.
6. Mirror the new sample to web.

   python3 web/samples/mirror_from_ios.py

   Same script used for the bundled-samples sync above. Copies the new .wav from the iOS folder into web/samples/Developer/ so it appears in the Bundled ▾ picker on web. (Web built-in presets don't yet auto-load bundled samples; for parity, the easiest path is to add a synth-only mirror entry to web/js/music.js in the same Developer Patches section.)

7. Commit + push. One commit covers all four touched areas: JGPresets/<name>.dronepreset, DroneMeditations/Samples/Developer/<name>.wav, the regenerated _summary.json / _generated_presets.swift, and the Preset.swift paste.

Regenerating the preset reference page — for maintainers + contributors

The page at preset-reference.html is auto-generated from DroneMeditations/Models/Preset.swift and web/js/music.js. Re-run the generator any time you add a preset, swap a sample, or tweak a curated Drone Artist note:

python3 tools/render_preset_reference.py

Hand-written artistic context for the Drone Artists category lives in tools/drone_artists_descriptions.json — keyed by preset name, value is a one-to-three-sentence HTML snippet rendered above each preset's auto voice table. Add or edit entries there and re-run the script to surface them on the page; other categories stay purely auto-described.

Upload your own

• Click Load file… on any oscillator's sample row.
• Pick a WAV / MP3 / OGG / FLAC file. Mono or stereo, any sample rate.
• The file is stored in your browser's IndexedDB (web) or the app's Documents folder (iOS) so it persists across sessions and into saved presets.

Sample voice behavior

• The frequency slider becomes a playback rate control — sliding up resamples the file faster (higher pitch).
• The same FX chain (filter / drive / chorus / delay / reverb / LFOs / drift) applies to samples.
• Samples are referenced by ID in user presets, so multiple presets can share one file without duplication.

Sample play window — trim, loop, fade

Once a file is loaded, a WINDOW row (amber-tinted) appears under the sample row with four sliders that shape which slice of the file loops:

• START (0–100%) — where in the file looping begins. Skip past silence, bowing scrapes, or a recording's noisy intro.
• END (0–100%) — where the loop wraps back to START. Trim off the decay tail so the loop point lands somewhere musically useful.
• FADE IN (0–10 s) — each loop pass fades up over this duration. Set to ~0.5–2 s to soften the loop-start seam on percussive material.
• FADE OUT (0–10 s) — each loop pass fades down toward END. The classic trick: equal fade-in and fade-out at ~⅓ of the window length give you a continuous breathing loop with no audible seam, even on harsh source material.

The window + fade settings are saved with full presets, so a bundled Tibetan-bowl sample with a 4-second window + 1-second fades becomes a fixed instrument inside the preset rather than the raw recording. Many of the new (Granular) and faded Drone Artists presets rely on this — the Basinski tape-decay cycle, for instance, uses a tight window with long crossfades to suggest the original loop is wearing away.

Granular sampling — web + iOS

A GRAINY toggle button appears under the WINDOW row whenever a sample is loaded. When on, continuous playback is replaced by a Hann-windowed grain scheduler that reads slices around a user-set position. Combined with the existing GRAIN row (size / density / jitter / spread) plus two new sliders, this is full granular sampling:

• pos (0–100%) — centre position in the sample. Grains are pulled from around this offset. 0 = start of file, 100 = end. A slow drift on pos via a pitch-LFO target gives you scanning-over-time textures.
• scan (0–100%) — per-grain position jitter. 0 = freeze on the exact offset (the most static, cymatic result). 100 = pull grains from anywhere in the sample (most chaotic). Mid-range gives a softly-rumbling cloud.

The grain scheduler is the same one used by the granular pink-noise waveform, so size / density / jitter / pan-spread behave identically — including the new BPM-quantized density chip, so you can lock sample-grain timing to a musical division. The only thing that changes between sample-granular and noise-granular is the source: a recognizable sample instead of pink noise.

What you can build with this: a Tibetan bowl strike frozen on its sustain forever, shimmering. The first second of a vocal vowel held indefinitely. A long ambient field recording chopped into rain-like fragments. A piano sustain decaying into a glittering cloud. Basinski-style tape-decay drones from any source. The Developer Patches presets (§6) lean heavily on this — most use 1/8t or 1/16 BPM-quantized grain density over a slow scan to produce intricate rhythmic patterns from a single sustained sample.

11The pill row

Shortcuts to the app's modal features.

AUTOMATION — scheduled events (web + iOS, v1.1)

A per-patch sequencer. Schedule chord changes, fades, waveform / sample switches, level / mute changes, and LFO rate / depth moves at musical positions; the dispatcher fires each one as the transport reaches it.

Tap the AUTOMATION pill (right after PRESET) to open a sheet listing scheduled events in time order. Tap + to add an event, tap a row to edit it, swipe left to delete. Each event has these fields:

• Position — when it fires, entered in bars + beats at the session BPM (e.g. "Bar 2, Beat 2.5" = the upbeat of beat 2 in the second bar), snapped to a 1/16-bar grid. Positions are tempo-relative: change the BPM and the whole timeline rescales in proportion, so the structure holds at any tempo.
• Apply to — All voices or OSC 1-4. All voices fires for every voice regardless of fade / mute state. A per-voice chord change transposes just that voice; All voices also updates the patch's key + chord pill.
• Action — one of eight: Chord change, Fade in / Fade out (0-15 s ramp), Waveform (including picking a bundled sample), Level (0-100% set), Mute toggle, plus LFO rate and LFO depth — target any of the 5 LFOs to, say, ramp an S&H pitch envelope from a whisper to a storm across a session.

Chord changes transpose, they don't respell. A chord event shifts each affected voice's pitch by an interval relative to the patch's starting state, so non-triadic voicings — Sunn O))) power chords, Solfeggio frequency stacks — keep their shape; only the root moves. A Direction toggle (Up / Down / Nearest) picks whether the move climbs, falls, or takes the shortest path. Each chord holds until the next chord event. The Duration picker is an editing convenience — it auto-positions the next event you add (so "1 bar" drops the next chord one bar later), letting you lay down a flowing progression in a few taps.

The Timeline section at the top sets Length (Manual stop, or 1-64 bars) and Repeat (Once → 16× → Forever). Manual stop = events fire once and the last state holds until you tap Stop. A bar length loops the phrase back to the start; Repeat caps how many times it runs. Every replay restores the patch's captured baseline first, so loops and Stop→Play always start clean — no drift across cycles.

Saved setups. The Saved setups section at the bottom of the sheet is a reusable library — tap Save current… to store the timeline under a name, then load it onto any patch later (independent of the sound). Handy for reusing a favourite chord progression or S&H build across different drones. Setups are stored per-device on each platform; a full patch's timeline still travels with it via .dronepreset.

DRIFT — generative slow motion

Pick a drift scene and the app gently wanders the chosen parameters over minutes. Scenes include:

• Glacial up / down — every voice slowly climbs or descends by a few cents.
• Octave up / down — voices migrate ±1 octave over the session.
• Wave / Ocean — sine-shaped pitch sway that breathes up and down forever. Wave is the slow up-and-back swing tied to the session timer; Ocean is a perpetual ~90-second ±¼ semitone tide that never resolves — the closest thing to a "still" preset that still feels alive.
• Opposite pan — voices 1&2 drift left, 3&4 right.
• Coordinated scenes — 2 voices up + 2 down, centered drone w/ pair drifting, full panning spiral, …
• Custom — set the dropdown on any single oscillator strip (🌀) to mix-and-match modes per voice.

Drift speed is calibrated to your session length: the full arc completes by the time the timer ends — unless you override it.

Per-voice drift overrides (Amount + Period)

Each oscillator strip's 🌀 Drift menu has four sections: Pitch (the motion shape), Amount (how far it travels), Period (how long one cycle lasts), and Pan (the stereo motion). Defaults follow the scene; pick a chip in Amount or Period to override that voice individually.

• Amount — ½ · 1 · 2 · 3 · 5 · 12 · 24 semitones. Lets you turn a glacial 12-cent drift into a 1- or 2-octave migration, or pull a 1-semitone bend down to a near-static ½-semitone breath.
• Period — 10 s · 30 s · 1 / 3 / 5 / 10 / 20 min (Ocean defaults to ~90 s if you don't override). Slower periods turn drift into pure ambience; faster periods make it audibly cyclic.
• An Auto chip in each section resets that voice to the scene's default amount/period.

These overrides make every voice's motion independent of session length, so you can build slow-bloom textures inside a long meditation: a 20-minute glacial-up bass paired with a 30-second wave on a high pad creates intentional layered motion that the simple "scene" presets can't produce alone.

Quantize to scale

At the top of each oscillator's 🌀 Drift menu there's a Quantize to scale toggle. When on, that voice's final pitch — drift + any pitch-targeted LFOs + FM, all combined — is snapped in real time to the nearest note of the current chord across roughly 2 octaves around the voice's base frequency.

What this turns into musically: continuous pitch wobble becomes arpeggio-like jumps along the chord tones. A slow wave drift that would normally smear ±¼ octave now steps through the chord notes inside that range. Faster LFOs on pitch turn into bouncing arpeggios. The voice always lands on chord-correct pitches, so dense modulation no longer dissonates against the rest of the chord.

Per-voice — so you can leave bass continuous and quantize only the high pad, or vice versa. Changes to the chord (CHORD pill, root, type, octave) recompute the snap targets immediately. Set to off to return to smooth/continuous drift.

LISTEN — tune to room

Pop the sheet, grant mic access, and play a sustained note from any instrument (voice, tuning fork, singing bowl). The app runs YIN pitch detection (de Cheveigné & Kawahara) in real time, displays the nearest note + cents offset + Hz + live audio level, and a "Set as Root" button snaps the chord generator's root key to match.

The displayed pitch is held on screen until you tap Reset or sing a new stable pitch — you don't have to race the readout to hit Set as Root. A subtle "HELD" badge appears when the value is sticky (mic went quiet). The status line reads "Held — tap Set as Root, or sing a new note".

On web there's a source picker dropdown — switch to a USB mic / interface input mid-session. iOS uses the system audio input.

Per-voice equivalent. If you want to tune only one oscillator (e.g. dial OSC 2 to whatever you're humming without retuning the others), use the 👂 ear icon on that strip instead — see §3 → 👂 Tune to Room. The global LISTEN pill on this page retunes the chord root, which means all four voices move together.

PERFORM — cymatics fullscreen

Hides every control element, leaving only the live Chladni pattern (with sand particles + zoom). A tiny "Exit" pill stays in the top-left; on iOS it auto-dims after 3 s but never disappears, and tapping anywhere wakes it back to full brightness. Esc exits on web.

Sand particles. Hundreds of drifting grains (~3000 on web, ~1500 on iOS) ride the cymatic gradient toward nodal lines while the Chladni pattern fills the screen — the same particle simulation as the pop-out window, now native to fullscreen. The grains physically migrate to new positions when frequency changes, so a slow pitch-LFO produces visibly sweeping sand currents. Turn the grains off with the sand toggle if you want a pure shader-only image.

JOURNEY — scripted meditations

25 multi-stage timed journeys auto-advance through presets + drift scenes. 15 general + 10 drone-artist:

• General: Sundown, Awakening, Floating, Body Scan, Cathedral, Mountain Climb, Vespers, Crystal Cave, Phi Spiral, Quartz, Lullaby, Tibetan Bowl, Storm Front, Centering, Spiral Descent.
• Drone-artist arcs: Deep Listening Lineage (Oliveros → Radigue → Stars of the Lid · 45 min), Heavy Resonance (Sunn O))) → Earth → Niblock · 30 min · drop master volume!), Minimalist Arc (Riley → Budd → Palestine · 25 min), Spiritual Path (Coltrane → Wada → Haino · 35 min), Sonic Cathedral, Black Mass, Slow Bloom, Tape & Tar, Awakening Drone, Microtonal Garden.

The session timer auto-sets to the journey total. Stop halts both the journey scheduler and the transport.

Compose your own. The "＋ Create your own journey" button at the top of the Journey sheet opens a composer where you name a journey and add any number of stages — each one a preset + drift scene + duration (0.5–90 minutes). User journeys are saved locally and appear in a "Your journeys" section above the curated list, with pencil (edit) and × (delete) controls.

MORPH — interpolate between two presets

The MORPH pill opens a sheet with two preset dropdowns and a 0–100% slider. As you drag, every per-voice parameter is interpolated between A and B in real time:

• Log interpolation on frequencies, filter cutoff & Q, reverb decay, delay time, chorus rate, FM index (musically smoother than linear across octaves).
• Linear on mix / depth / pan / drive / feedback / width.
• Discrete swap at the midpoint (≥ 50%) on waveform, filter type, delay mode, FM source, LFO shape + target.

The preset name flips to "A → B (NN%)" while morphing. Clear morph wipes the From/To pair and returns to plain preset behavior.

Auto-morph (time-driven). Below the slider, the Auto-morph panel lets you pick a duration (30 s · 1 / 3 / 5 / 10 / 20 / 30 / 60 min) and tap ▶ Play. The morph slider then crawls from 0 → 100 % over that duration, applying every parameter blend continuously — a 20-minute Oliveros → Basinski morph becomes a guided meditation in itself. ⏸ Pause freezes mid-morph; ↺ Reset snaps back to 0 % without losing the From/To selection. Tick Ping-pong to bounce back and forth between A and B forever. The timer keeps running when you close the sheet, so you can dismiss the controls and watch Chladni evolve through the full duration. Especially fun between Drone Artists presets — try morphing Oliveros → Stars of the Lid, or Sunn O))) → Earth.

GALLERY — cymatics snapshots (web)

Every snapshot you save in the pop-out Chladni window is mirrored here. Tap a thumbnail to re-download; the × button removes it from the gallery only — already-downloaded PNGs aren't touched. Capped at 200 to keep localStorage healthy.

12Chladni visualization

A physically-calibrated render of how each frequency would vibrate a square plate.

The field is drawn per-pixel on the GPU on both platforms — a WebGL fragment shader on the web and a Metal shader on iOS — so the nodal lines stay sharp and animate smoothly with vibrato at any zoom, at low CPU cost. Sand grains ride on top.

The frequency → pattern mapping was fit from 17 brusspup demo frames (345 Hz → 6051 Hz), where each frame was visually identified as a (m,n) antisymmetric eigenmode on a thin square plate:

f(m, n) ≈ K · (m² + n²), K ≈ 18.6 ± 0.8 Hz

For each voice, the renderer picks the two adjacent eigenmode pairs that bracket the live frequency and crossfades between them, so vibrato (pitch-LFO modulation) breathes smoothly between physical modes instead of snapping or following an arbitrary continuous curve.

The Chladni renderer also adds a small center "driver bolt" (brusspup's plate is center-driven), and a thin nodal ring at small radius, on top of every frame.

Tiling

The Chladni cos math is naturally periodic, so the pattern tiles to fill the viewport at any zoom level. You'll never see a clipped plate boundary — even zoomed all the way out, the whole screen is filled.

Zoom

The vertical zoom slider on the left (web) or pinch gesture (iOS) controls plate scale, 0.25× to 4×. Persists across launches. On iOS the slider/pinch fades when the main controls are visible so it never competes with the panel.

13Pop-out Chladni (web)

A dedicated window with WebGL-accelerated rendering, 3000 sand particles, and snapshot export.

• Open from the Chladni toggle's pop-out arrow.
• Receives live oscillator state from the main window via BroadcastChannel — frequencies, mutes, solos, vibrato all sync at ~15 fps.
• 3000 sand grains drift along the gradient toward nodal lines. Particles physically migrate to new positions when frequencies change.
• Bottom controls strip mirrors frequency/solo/mute for each voice — adjust without flipping back to the main window.
• The 📷 snapshot button (or S) saves a PNG to Downloads and mirrors it into the in-app Gallery.
• The sand toggle (or G) removes the particle layer for a cleaner shader-only image.
• Scroll wheel zooms; Esc closes the window.

14Spectrum analyzer

Real-time FFT view tapped post-limiter.

Toggle the chart-bar button in the header to overlay a log-frequency spectrum bar graph (20 Hz – 16 kHz). It's a quick way to verify what's actually leaving the master bus — useful when chasing why a "bass-only" preset isn't producing low energy (often a runaway high-pass filter or an LFO killing the fundamental).

The web version uses AnalyserNode.getByteFrequencyData; iOS uses vDSP forward FFT with a Hann window on a tap installed on engine.outputNode.

15Transport & recording

Play / pause / stop

The round play button fades audio in over ~1 s (or 3 s if this is the first play of a fresh session). Pause and Stop both do something different from any other synth's transport — read on.

Reverb-bloom fades (iOS)

When you tap Pause or Stop, the app doesn't just drop the master volume. Each voice's reverb mix and decay simultaneously bloom upward, then descend back to your preset values as the master volume fades to silence. The effect is the sound dissolving into the room rather than getting quieter and quieter — like the last note of a song letting the hall finish for it.

• Stop: 8 s total fade. Logarithmic master descent (every second drops the same number of dB, so the wind-down feels uniform). Reverb bloom peaks at ~2.4 s (mix climbs to 0.7, decay to 7 s), then ramps back down so the wet signal joins the master in silence by t=8 s.
• Pause: 1.4 s total fade. Same triangular shape, smaller bloom (peak mix 0.35, decay 3.5 s, peaks at ~0.35 s). A gentle "lift then settle" — quick enough to feel like a pause, atmospheric enough to feel composed.

Both fades restore each voice's original reverb settings on completion (or immediately, if you press Play while a fade is in flight) — so your preset reverb mix isn't permanently changed. Stop also resets the session timer to 0.

Session timer

Default 30 min. Tap the timer readout to change the duration. The app fades out gently in the last few seconds. Useful when you're using Drone Meditations as a sleep aid — set 20 min and trust it to fade itself.

BPM & metronome

The tempo readout and the metronome toggle live in the top icon row, always visible (they used to be buried in the master strip — awkward to reach in iPhone landscape). Tap the BPM to set the session tempo; tap the metronome icon for an audible click locked to the transport's downbeat.

BPM is the master clock for everything tempo-aware: delay-time sync, LFO rate-sync, granular density sync, and the Automation Timeline (bar/beat positions resolve at the live BPM, so changing tempo rescales a whole automated phrase in proportion).

Recording (web + iOS)

The record button writes the master output to a file:

• Web — WebM/Opus or WebM/PCM (whatever your browser supports), saved to Downloads.
• iOS — two files per session, both written into Drone Meditations/Recordings/ in the Files app:
  • .wav — 24-bit uncompressed PCM. The primary deliverable: bit-perfect, professional-grade, ideal for editing in a DAW or archival. Large (~17 MB/min stereo).
  • .m4a — AAC sidecar, ~10× smaller, identical loudness + fades baked in. Ideal for AirDropping, emailing, or dropping into the Music app.
  Both files are mastered through the same pipeline: a single gain stage targets ≈ -16 dBFS RMS while keeping peaks below -0.3 dBFS, then a 2-second linear fade-in and 4-second fade-out are applied so the file doesn't start or end with a click. The M4A also gets embedded title / artist / date metadata so it shows up nicely in Music and Files.

Recording captures whatever is going through the master limiter — so all FX, drift, LFO modulation, journeys, etc. — exactly as you hear them. The share sheet appears automatically when mastering finishes (typically < 1 second after Stop).

16MIDI input (web)

Any connected MIDI controller can drive the chord generator's root key.

Connect a USB MIDI keyboard / controller and press any key. The chord generator's root key snaps to the played note (octave clamped to the 1–6 range); the active chord template stays the same so you can play the chord type by sliding around the keyboard.

Web MIDI is initialized on first user gesture (click anywhere). Browsers that don't support Web MIDI (Safari without flags) silently fall back to no-MIDI.

17iOS-only features

First-launch tour

The first time you open the iOS app, a 5-card tour walks you through Play / Preset / Tweak / Cymatics / Go Further. Swipe through or tap Next; Skip jumps to the synth. You can re-open the tour any time from the ? button in the controls header.

Now Playing / Lock Screen

While playing, iOS shows Drone Meditations on the Lock Screen and Control Center with play / pause controls. Plays through silent mode (the audio session uses the .playback category) and supports background audio so you can lock the screen mid-session.

Haptics

Optional. The haptic button in the master row cycles through three states on tap: Off → Light → Heavy → Off. The device taps in sync with the slowest active LFO across all voices; depth determines per-tap intensity, and the mode scales it: Light halves the intensity for a barely-there pulse, Heavy uses the full computed scale. Most noticeable with a 0.5–1 Hz amplitude LFO. The choice persists across launches.

Photos library snapshot

The 📷 camera button renders a 1500×1500 frame of the current Chladni state and saves it directly to your Photos library (add-only permission — the app never reads your existing photos). First save prompts for permission; subsequent saves are silent. A "Saved to Photos" toast confirms.

Pinch zoom

Two-finger pinch on the Chladni view zooms in and out. Persists across launches.

Performance exit safety

In cymatics fullscreen, the tiny "Exit" pill dims to 45% after 3 seconds, but never fully disappears, and tapping anywhere on screen brings it back to full brightness. (You can't get stuck.)

18Keyboard shortcuts (web)

19A note on the Solfeggio claims

An honest, evidence-based take.

The Solfeggio preset descriptions in this app use language like "Traditionally associated with…" — and that wording is deliberate.

The popular Solfeggio system (174, 285, 396, 417, 528, 639, 741, 852, 963 Hz) is a 20th-century invention. It was published in 1999 by Joseph Puleo and Leonard Horowitz; the frequencies were derived numerologically (each reduces to a single digit in a specific Pythagorean scheme), not from any medieval chant or ancient text. The medieval "ut/re/mi/fa/sol/la" was a solmization system, not a list of frequencies.

There is no peer-reviewed evidence that 528 Hz "repairs DNA," that 396 Hz "liberates guilt," or any of the other claims that circulate around these tones. A handful of papers in low-rigor journals (including predatory or low-impact outlets) discuss possible psychoacoustic effects, but these don't replicate at the level mainstream medicine would accept.

That said: specific frequencies do affect mood, attention, and physiology — and so does sitting still for 20 minutes listening to anything calming. The presets are included because people enjoy them and the harmonic context is interesting, not as a medical claim. If you find them useful for relaxation or focus, that's a real and welcome outcome — just don't expect the literal claims attached to them in pop wellness culture.

20Tips & gotchas

21Credits

Drone Meditations is a personal project by Jose Gude MD, designed and shipped on both web and iOS in 2026.

Special thanks to:

• brusspup (Eric Roner) — for the Chladni demo footage used to physically calibrate the visualization.
• The Web Audio, AVAudioEngine, and SwiftUI teams — for primitives stable enough to ship a 4-voice synth in a weekend.
• Wendy Carlos, Harry Partch, Lou Harrison — for the tuning systems that broaden what "in tune" can mean.

← Back to Drone Meditations