Strudel Music

Strudel Music 🎵

Like /imagine for Midjourney, but for music. Compose, render, and stream music using Strudel live-coding patterns — entirely offline, no browser required.

Command Interface

When a user invokes /strudel, route based on intent:

/strudel <prompt> — Compose from description

Generate a Strudel pattern from a natural language prompt. The agent interprets the mood, instruments, tempo, and structure, writes a composition file, renders it to audio, and posts the result.

Examples:

• /strudel dark ambient tension, low drones, sparse percussion, 65bpm
• /strudel upbeat tavern music with fiddle and drums
• /strudel lo-fi chill beats to study to
• /strudel epic battle music, brass and timpani, 140bpm
• /strudel a theme for a character named Cael — curious, quick, a little dangerous

Workflow:

1. Parse prompt → select mood, key, tempo, instruments from the decision tree
2. Write a .js composition file using Strudel pattern syntax
3. Render via node src/runtime/offline-render-v2.mjs <file> <output.wav> <cycles> <bpm>
4. Convert: ffmpeg -i output.wav -c:a libmp3lame -q:a 2 output.mp3
5. Post the MP3 as an attachment in the channel
6. Optionally: play in Discord VC if user is in a voice channel

/strudel play <name> — Play in Discord VC

Stream a saved composition into the user's Discord voice channel.

# Render + convert + stream
node src/runtime/offline-render-v2.mjs "assets/compositions/<name>.js" /tmp/<name>.wav 16 120
ffmpeg -i /tmp/<name>.wav -ar 48000 -ac 2 /tmp/<name>-48k.wav -y
node scripts/vc-play.mjs /tmp/<name>-48k.wav

/strudel list — Show available compositions

List all .js files in assets/compositions/ with their metadata (@title, @mood, @tempo).

/strudel samples — Manage sample packs

Subcommands:

• /strudel samples list — show installed sample directories and counts
• /strudel samples download — re-run scripts/download-samples.sh (idempotent — skips if already present)
• /strudel samples add <url> — download a sample pack from a URL (ZIP/tar containing WAV directories)
• /strudel samples add <path> — symlink or copy a local directory into samples/

How custom samples work: Drop any directory of WAV files into samples/<name>/. They're automatically discovered by the renderer. Use them in patterns with s("<name>"). Variations are indexed by filename sort order — access with s("<name>").n(3).

Example: if you have an Ableton drum rack exported as WAVs:

samples/
  my-kit/
    kick-soft.wav    → s("my-kit").n(0)
    kick-hard.wav    → s("my-kit").n(1)
    snare-tight.wav  → s("my-kit").n(2)
    snare-loose.wav  → s("my-kit").n(3)

/strudel concert <name> [name2] [name3] ... — Play a setlist

Render and stream multiple compositions sequentially into Discord VC.

Setup

npm run setup
# Installs all deps + downloads dirt-samples (~11MB, CC-licensed)

That's it. First render: npm run test:render

Adding more sample packs

The skill ships with dirt-samples (96 WAVs: kicks, snares, hats, toms, 808s). For richer sounds, add sample packs:

CC0 / Free packs (just download and drop in samples/):

• Dirt-Samples — 800+ samples (full pack, we ship a subset)
• Signature Sounds – Homemade Drum Kit (CC0) — 150+ one-shots
• Looping – Synth Pack 01 (CC0) — synth one-shots + loops
• artgamesound.com — CC0 searchable aggregator

Your own packs: Export from any DAW (Ableton, FL Studio, M8 tracker, etc.) as WAV directories. Strudel doesn't care where they came from — it's just WAV files in folders.

Named banks (Strudel built-in, requires CDN access):

sound("bd sd cp hh").bank("RolandTR909")
sound("bd sd hh oh").bank("LinnDrum")

WSL2 Note

If running on WSL2 and streaming to Discord VC, enable mirrored networking:

# %USERPROFILE%\.wslconfig
[wsl2]
networkingMode=mirrored

Then wsl --shutdown and relaunch. Without this, WSL2's NAT breaks Discord's UDP voice protocol — the bot joins the channel but no audio flows because IP discovery packets can't traverse the NAT return path. Mirrored mode eliminates the NAT by putting WSL2 directly on the host's network stack.

This only affects VC streaming. Offline rendering and file posting work in any networking mode.

How It Works

The offline renderer uses node-web-audio-api (Rust-based Web Audio for Node.js) for real audio synthesis:

1. Pattern evaluation — @strudel/core + @strudel/mini + @strudel/tonal parse pattern code into timed "haps"
2. Audio scheduling — Each hap becomes either:
   • An oscillator (sine/saw/square/triangle) with ADSR envelope, biquad filter, stereo pan
   • A sample (AudioBufferSourceNode) from the samples directory, with pitch shifting
3. Offline rendering — OfflineAudioContext.startRendering() produces complete audio
4. Output — 16-bit stereo WAV at 44.1kHz → ffmpeg → MP3/Opus

Note on mini notation: The renderer explicitly calls setStringParser(mini.mini) after import because Strudel's npm dist bundles duplicate the Pattern class across modules. Same class of bug as openclaw#22790.

Composition Reference

Tempo

setcpm(120/4)  // 120 BPM

Layering

stack(
  s("bd sd bd sd"),                              // drums
  note("c3 g3").s("sawtooth").lpf(800),          // bass
  n("0 2 4 7").scale("C:minor").s("triangle")    // melody
)

Pattern syntax

• "a b c d" — sequence (one per beat)
• "[a b]" — subdivide (two in one beat)
• "<a b c>" — alternate per cycle
• "a*3" — repeat
• "~" — rest/silence
• .slow(2) / .fast(2) — time stretch
• .euclid(3,8) — Euclidean rhythm

Expression

.lpf(sine.range(400, 4000).slow(8))   // filter sweep
.gain(sine.range(0.1, 0.2).slow(9))   // breathing volume
.pan(perlin.range(0.2, 0.8))          // organic stereo
.room(0.5).roomsize(4)                 // reverb
.delay(0.3).delaytime(0.25)           // delay
.attack(0.01).decay(0.2).sustain(0.5).release(0.3)  // ADSR

Song structure

let intro = stack(pad, noise)
let verse = stack(drums, bass, melody)
let chorus = stack(drums, bass, melody, lead)

arrange(
  [8, intro],
  [16, verse],
  [8, chorus]
).cpm(120/4)

Mood→Parameter decision tree

See references/mood-parameters.md for the full tree with transitions and leitmotifs. See references/production-techniques.md for advanced techniques (breathing, shimmer, earth pressure).

File Structure

src/runtime/
  offline-render-v2.mjs    — Core offline renderer
  smoke-test.mjs           — 12-point verification test

scripts/
  download-samples.sh      — Download dirt-samples (idempotent)
  vc-play.mjs              — Stream audio to Discord VC

samples/                   — Sample packs (gitignored, downloaded on demand)
assets/compositions/       — Saved compositions
references/                — Mood trees, techniques, architecture docs