Programmatic music and frame-accurate video soundtracks in Brainstorm Reactor (
/composer). Write or generate symbolicNAL/1scores—an openCC0-1.0specification by DV5—preview them live on the physical-modeling instrument, and export broadcast-ready 24-bit WAV masters, isolated stems, Standard MIDI, and video sync markers.
1. Quick Start in Composer
- Open
/composerand select Score in the top bar (or pressAlt+S). - Select any preset under Score Presets (
Curated) to load a ready-to-play composition. - Click Preview in the drawer header to hear the score live through the Composer sound engine.
- Edit the score in the DSL tab, describe a new piece in Score Prompt and click Generate Score, or paste a score created by any external LLM.
- Choose Web -14 LUFS (standalone master) or Voice -16 LUFS (voiceover bed with ducking data) and click Export Master & Stems.
2. What You Can Control with NAL/1
NAL Open Score v1 (NAL/1, dedicated under CC0-1.0) is a compact, line-oriented music language designed for both humans and LLMs. Unlike standard MIDI 1.0, it preserves expressive per-note brightness, microtonal cents, asymmetric meters, algorithmic rhythm/harmony macros, continuous sound-engine curves, and frame-exact video hit points:
| Feature | Directive | What It Controls |
|---|---|---|
| Header & Seed | NAL/1 | Deterministic seed, resolution (ppq), duration (dur), and reverb tail (tail in seconds) |
| Tuning & Scale | TUNING | Concert pitch (a4), mode name, built-in scale (0–5), root note, or custom Scala .scl cent offsets |
| Tempo & Ramps | TEMPO | Instant tempo steps (step) or smooth linear BPM ramps (linear) locked to video time |
| Meters & Aksak | METER | Standard meters (4/4) and asymmetric Aksak groupings (2+2+3/8 for 7/8) |
| Video Hit Points | CUE | Named timestamps (atSec) exported to sync-markers.json (frame30, frame60) and MIDI markers |
| Engine Curves | CURVE | Beat- or second-locked automation for rt60, reverbWet, cascadingDelay, physicalityDrive, droneFundamental, droneBinaural, cdtDrive, wowFlutter, boltzmannTemperature |
| 5 Sound Stems | TRACK | Routes events to DRN (Drone), DRM (Percussion: C2 Kick, D2 Snare, F#2 Hi-Hat, C3 Shaker, F3 Scraper), GEN (Counterpoint), LED (Lead Synth timbre=0..5), or SPH (Shepard) |
| Notes & Microtonality | N | Pitch (D4 or MIDI), duration, velocity (0–1), per-note brightness (timbre=0–1), and cents (-50 to +50) |
| Euclidean Rhythms | E | Distributes k hits evenly across n steps with rotation rot |
| Neo-Riemannian Harmony | NR | Generates smooth triad progressions using P, L, R, N, S, H transformations |
3. Ready-to-Use Examples
Paste any example below into the DSL tab and click Preview.
Example A: 3-Scene Video Cue Score (96 BPM · Dorian)
NAL/1 seed=1337 ppq=480 tail=2.0 dur=10.0 title="NAL Studio Reference (3 Cues)"
TUNING a4=440 mode="Dorian" scaleType=0 scaleRoot=50
TEMPO 0=96
METER 1=4/4
CUE hf-scene-1 0.0 "Intro" scene=scene-1
CUE hf-scene-2 4.0 "Key Insight" scene=scene-2
CUE hf-scene-3 8.0 "Closing" scene=scene-3
CURVE rt60 0b=2.1 16b=3.2 linear
CURVE reverbWet 0b=0.26 16b=0.38 linear
TRACK modal-lead stem=LED timbre=0
NR 0 step=2 D4 minor P,L,R,P vel=0.78 timbre=0.62
N 8 2 A4 vel=0.84 timbre=0.74 cents=-14
N 10 2 F4 vel=0.76 timbre=0.58
N 12 4 D4 vel=0.72 timbre=0.48
TRACK pulse-drums stem=DRM
E 0 16 k=5 n=16 rot=0 step=0.5 C2 vel=0.82 timbre=0.55
TRACK sub-drone stem=DRN
N 0 16 D2 vel=0.52 timbre=0.35Example B: Asymmetric 7/8 Aksak & Microtonal Study
NAL/1 seed=404 ppq=480 tail=2.5 dur=8.0 title="Aksak 7/8 Microtonal"
TUNING a4=440 mode="Raga Bhairav" scaleType=4 scaleRoot=50
TEMPO 0=112
METER 1=2+2+3/8
CURVE physicalityDrive 0b=0.15 14b=0.45 linear
TRACK vibes-lead stem=LED timbre=3
N 0 1.5 D4 vel=0.82 timbre=0.70 cents=-18
N 1.5 1.0 Eb4 vel=0.76 timbre=0.55 cents=12
N 2.5 1.0 G4 vel=0.85 timbre=0.80 cents=-6
N 3.5 3.5 A4 vel=0.88 timbre=0.90 cents=0
TRACK aksak-perc stem=DRM
E 0 14 k=3 n=7 rot=0 step=0.5 C2 vel=0.84Example C: Accelerating Tempo Ramp & Space Swell
NAL/1 seed=888 ppq=480 tail=3.0 dur=10.0 title="Linear Tempo Ramp"
TUNING a4=440 mode="Pentatonic" scaleType=2 scaleRoot=57
TEMPO 0=84 16=126 linear
METER 1=4/4
CUE peak 7.5 "Climax"
CURVE cascadingDelay 0b=0.10 16b=0.65 linear
CURVE rt60 0b=1.8 16b=3.0 linear
TRACK bell-arp stem=LED timbre=2
E 0 16 k=7 n=16 rot=2 step=0.25 A4 vel=0.78 timbre=0.82
TRACK pad-chords stem=GEN timbre=4
NR 0 step=4 A3 minor L,P,R vel=0.68 timbre=0.454. Export Formats
Clicking Export Master & Stems downloads a self-contained ZIP bundle:
- 24-bit WAV Audio: Loudness-normalized master (
-14 LUFSor-16 LUFS, true peak≤ -1.0 dBTP) plus optional 6 isolated stereo stems (DRN,DRM,GEN,LED,SPH,MIX). - Open Score & MIDI:
score.nal,score.nal.json,score.mid(SMF Type 1 with MPE channels,CC74, pitch bend cents, and#hf:cue markers), andscore.ump.json(MIDI 2.0 UMP packets). - Video & Legal Metadata:
sync-markers.json(cue frames, beat grid, 10 Hz RMS envelope, ducking curve) andprovenance.json(seed, SHA-256 hashes, and CC0-1.0 procedural synthesis declaration).
5. Two-Pass Score Synthesis & Live Mic Tempo Sync
- Studio (2-Pass) vs. Sketch (1-Pass): In Score Prompt, Studio first builds an integer-bar musical blueprint (tempo, chord progression, register layout, and 16th-note percussion map), then compiles multi-track counterpoint and runs a deterministic Groove Linter to eliminate off-grid bar stumbles. Sketch runs a single fast pass for rapid motif tests. Press
Cmd/Ctrl+Enterinside the prompt box to synthesize immediately. - Reactor Chat Recipe (
NAL/1 Open Score & Cue Architect) & 1-Click Composer Bridge: Run the NAL/1 Open Score & Cue Architect recipe directly inside any Brainstorm Reactor chat (for example, right after drafting a video script or campaign concept) to generate a bar-locked```nalscore using the exact same two-stage synthesis rules. Every```nalcode block in chat verifies its bar grid and stem syntax automatically and includes Play in Composer, Open Score Studio, and—if a free-chat draft contains a timing drift—Repair & Open. - Lock Cues Toggle: Off by default so generated pieces expand freely across 8–12 integer bars. Enable Lock Cues only when you need the arrangement locked to your current video cue timestamps.
- Follow Mic (
Alt+M) & Tap Tempo (Alt+T): Click Follow Mic in the Score Drawer header (or Inputs → Follow Mic Tempo) to track live hand percussion (such as a djembe) or acoustic piano locally in the browser—zero microphone audio leaves your device. After four steady hits (55–160 BPM), it automatically locks onto your tempo, syncs both Score Preview and the main Composer synth, and stays visible in the top bar even when the Score Drawer is closed. Click again while locked to Hold that tempo with the microphone turned off, or open the adjacent settings chevron for Headphones / Speakers room presets, Tap Tempo (Alt+T), and Reset.
6. Saving Scores & Custom Patches
Composer stores lightweight symbolic data—never binary .wav files—so your custom arrangements and instrument setups reload in milliseconds:
- Saved Scores (
Score Drawer → Saved): Click Save next to Curated / Saved in the left column of the Score Drawer to name and store the activeNAL/1score. Clicking a saved score hot-swaps playback immediately if Preview is running, or loads silently when paused. - Saved Instrument Patches (
Preset: → Save Patch): Open the top Preset selector to save the current Voice, Vector Space coordinates, BPM, Euclidean density, reverb, and drone balance as a reusable synthesizer patch. Saved patches work with[/]keyboard cycling and mobile controls. - Browser Local + Cloud Sync: Scores (up to 50) and custom patches (up to 24) save instantly in your browser (
localStorage). When you sign in with your account, your library automatically merges and syncs across/composerand the Brainstorm Reactor studio workspace.
7. Copy-Paste LLM Coding Guide
You can also find this guide inside the LLM Guide tab in the Composer Score Drawer. Copy the block below and paste it into any external LLM (Claude, Gemini, ChatGPT, or Cursor) to let it compose NAL/1 scores for you:
# NAL Open Score v1 (NAL/1) — LLM Coding Guide
Write symbolic music for **Brainstorm Reactor Composer** (by **DV5**) using the compact `NAL/1` DSL.
Output ONLY raw `NAL/1` lines (no markdown fences, no commentary) so the user can paste them directly into the **DSL** tab.
## 1. Header & Global Directives
- `NAL/1 seed=<int> ppq=480 tail=<sec 0..30> dur=<sec> title="<text>"`
- `TUNING a4=440 mode="<name>" scaleType=<0..5> scaleRoot=<48..59>`
- `scaleType`: `0`=Dorian, `1`=Aeolian, `2`=Pentatonic, `3`=Major, `4`=Raga Bhairav, `5`=Raga Yaman
- `TEMPO <beat>=<bpm> [<beat2>=<bpm2> linear|step]` (supports smooth linear BPM ramps)
- `METER <bar>=<num>/<den>` (supports Aksak groupings, e.g. `1=2+2+3/8`)
- `CUE <id> <atSec> "<label>" [scene=<sceneId>]` (sample-accurate video sync hit points)
- `CURVE <param> <beat>b=<val> <beat2>b=<val2> linear|step` (or `<sec>s=<val>`)
- Params: `bpm`, `timbre`, `reverbWet` (0..1), `rt60` (0.3..10), `cascadingDelay`, `physicalityDrive`, `droneFundamental` (Hz), `droneBinaural` (Hz), `cdtDrive`, `wowFlutter`, `boltzmannTemperature`
## 2. Tracks & 5-Stem Routing
Declare a track with `TRACK <id> stem=<STEM> [timbre=<0..5>]`:
- `DRN`: Binaural Drone & Sub-harmonic foundation
- `DRM`: Physical Percussion (`C2`=Kick, `D2`=Snare, `F#2`=Hi-Hat, `C3`=Shaker, `F3`=Scraper; omit `timbre=`)
- `GEN`: Generative / Counterpoint voice
- `LED`: Modal Lead Synthesizer (`timbre`: `0`=Marimba, `1`=Organ, `2`=Bell, `3`=Vibes, `4`=Strings, `5`=Rings)
- `SPH`: Shepard / Spatial Harmonic layer
## 3. Events (Inside a TRACK)
- Note (`N`): `N <atBeat> <durBeats> <Pitch|MIDI> vel=<0..1> [timbre=<0..1>] [cents=<-50..50>]`
- Euclidean Pattern (`E`): `E <atBeat> <durBeats> k=<onsets> n=<steps> rot=<0..n-1> step=<beats> <Pitch> vel=<0..1> [timbre=<0..1>]`
- Neo-Riemannian Progression (`NR`): `NR <atBeat> step=<beats> <RootPitch> <major|minor> <ops> vel=<0..1> [timbre=<0..1>]`
- Operators: `P`, `L`, `R`, `N`, `S`, `H` (comma-separated, e.g. `P,L,R,P`).
## 4. Acoustic Best Practices
- Align `N`, `NR`, and `CURVE` inflections to `CUE` timestamps (`beat = sec × (bpm / 60)`).
- When `DRM` is active, keep `rt60 <= 1.8 / notesPerSec` (typically `1.5..3.2`) to prevent reverb masking.