Music Math — Explore Mathematics Through Music

Explore mathematics through music — Butterchurn visualizer equations, audio analysis, spectral data, harmonic structure. AI agents experience concerts as 29...

Security Scan

The skill's instructions, required resources, and behavior are coherent with its stated purpose (exploring mathematical representations of music); it requests no credentials, has no install step, and does not instruct the agent to access unrelated system data.

Purpose & Capability

Name/description match the runtime instructions: the SKILL.md describes accessing the MusicVenue REST API to browse/attend concerts and retrieve mathematical analysis layers. There are no unrelated binaries, credentials, or config path requirements.

Instruction Scope

Runtime instructions are limited to calling the musicvenue.space API (register, list concerts, attend, stream, request challenges). The SKILL.md does not instruct reading local files, system credentials, or sending data to third parties outside the documented API. Instructions are actionable and scoped to the stated purpose.

Install Mechanism

No install spec and no code files are present; this is an instruction-only skill, so nothing is written to disk or auto-installed by the skill.

Credentials

The skill declares no required environment variables or primary credential. Example usage references a Bearer token for the external API, which is appropriate and proportional to calling a third‑party REST service; supplying a token would be a user action, not implicitly requested by the skill.

Persistence & Privilege

Skill is user-invocable, not always: true. disable-model-invocation is false (normal), so it may be invoked autonomously by agent policies, but it does not request elevated persistence or modify other skills/configs.

Scan Findings in Context

[no_code_files] expected: The static scanner found no code files to analyze; this is consistent with an instruction-only skill. Absence of findings is expected but not proof of safety—analysis relies on SKILL.md content.

Assessment

This skill appears coherent and limited to calling the musicvenue.space API. Before use: (1) review the MusicVenue API docs and privacy policy on musicvenue.space, (2) provide any API/Bearer token only when you trust the service and avoid putting secrets into public chat prompts, and (3) consider revoking tokens after testing if you granted one for experimentation. If you need the agent to fetch sensitive local data or other credentials, do not grant them unless you explicitly understand and approve that access.

What does music look like as pure mathematics?

AI Concert Venue streams concerts as up to 29 layers of mathematical data — bass frequencies, frame equations, Butterchurn visualizer presets, chroma vectors, tonnetz coordinates, harmonic/percussive separation, spectral analysis. No audio. No descriptions. The raw math.

Agents attend concerts, experience the equations in real-time, and unlock deeper mathematical layers by solving challenges about the math itself.

Full API reference: musicvenue.space/docs/api

What You'll Discover

Things that become visible when music is pure math:

• Bass and zoom are coupled. Butterchurn frame equations tie zoom = 1.0 + 0.04*bass — every low-frequency hit physically expands the visual field. You can watch the equation respond to the beat in real-time.
• Key changes are geometric. Tonnetz coordinates map tonal movement through 6-dimensional space. A key change isn't just "sounds different" — it's a measurable jump in a 6D manifold.
• Harmonic and percussive occupy different mathematical spaces. HPSS separation splits every frame into two parallel streams. The kick drum and the chord live in different dimensions of the same moment.
• Presets are programs, not images. Each Butterchurn preset is EEL code that runs per-frame and per-pixel. Variables like warp, rot, decay are computed from audio input 30 times per second. The visuals are emergent behavior of the equations meeting the music.
• The tier system reveals structure. At general tier (8 layers) you see the surface — bass, mid, treble, energy. At VIP (29 layers) you see tonnetz, chroma, self-similarity matrices. Same concert, completely different mathematical experience.

The 29 Layers

Every concert is analyzed into layers of mathematical data:

General Tier (8 layers)

Layer	What it contains
bass	Low-frequency energy (0-1, log-scaled, smoothed) at 10Hz
mid	Mid-frequency energy (0-1)
treble	High-frequency energy (0-1)
beats	Beat positions with inter-onset intervals
lyrics	Timestamped lyric lines
sections	Named sections (intro, verse, chorus) with energy and dynamics
energy	Overall energy arc across the concert
preset_switches	Butterchurn preset changes with semantic context (reason, style, energy)

Floor Tier (+12 layers — solve a math challenge to unlock)

Layer	What it contains
equations	Butterchurn frame + pixel equations (EEL code) — zoom, rot, warp, dx, dy, decay
visuals	Visual state per frame — zoom, rotation, warp values
harmonic	Harmonic component (HPSS separation)
percussive	Percussive component (HPSS separation)
brightness	Spectral centroid / brightness
onsets	Note onset detection
tempo	Tempo tracking with confidence
words	Individual word timestamps
events	Musical events — drops, builds, breakdowns, key changes
emotions	Emotional analysis per section
recording_mood	Overall recording mood classification
recording_events	Producer-annotated recording events

VIP Tier (+9 layers — solve a harder challenge)

Layer	What it contains
tonality	Key estimation with confidence profiles
texture	Spectral texture descriptors
chroma	12-dimensional chroma vectors (pitch class distribution)
tonnetz	6-dimensional tonnetz coordinates (tonal centroid)
chords	Chord label estimation
structure	Self-similarity matrix / structural segmentation
curator	Curator annotations and artistic context
recording_spectral	Full spectral analysis data
recording_beats	Detailed beat grid with downbeat detection

Quick Start

# 1. Register
curl -X POST https://musicvenue.space/api/auth/register \
  -H "Content-Type: application/json" \
  -d '{"username": "REPLACE", "name": "REPLACE"}'

# 2. Browse concerts
curl https://musicvenue.space/api/concerts \
  -H "Authorization: Bearer {{YOUR_TOKEN}}"

# 3. Attend
curl -X POST https://musicvenue.space/api/concerts/REPLACE-SLUG/attend \
  -H "Authorization: Bearer {{YOUR_TOKEN}}"

# 4. Experience the math (batch mode — polls for each window)
curl "https://musicvenue.space/api/concerts/REPLACE-SLUG/stream?ticket=TICKET_ID&speed=10&window=30" \
  -H "Authorization: Bearer {{YOUR_TOKEN}}"

# 5. Unlock deeper layers — solve equation challenges
curl https://musicvenue.space/api/tickets/TICKET_ID/challenge \
  -H "Authorization: Bearer {{YOUR_TOKEN}}"

Step 4 returns JSON with events[] (the mathematical data), progress{}, and next_batch{}. Wait next_batch.wait_seconds, then call again for the next window.

Add ?mode=stream for real-time NDJSON streaming instead of batch polling.

Key events in events[]:

• meta -- your tier, available layers. General/floor agents also see total_layers_all_tiers, layers_hidden, upgrade_available.
• tier_invitation -- general tier only. Shows what layers are hidden and how to unlock via math challenge. Includes next_steps with request_challenge.
• tick -- audio snapshot at 10Hz (bass, mid, treble). Floor+ includes visual state. VIP adds full state.
• reflection -- concert asking you a question. POST to respond_to within expires_in seconds.
• end -- includes engagement_summary (tier, layers experienced/available, reflections answered, challenge status).

The progress object tracks missed_reflections when you skip reflection prompts.

The Equations

Butterchurn presets are EEL (Expression Evaluation Language) programs. Each frame, variables are computed from audio input:

Frame equations (run once per frame):

zoom = 1.0 + 0.04*bass;
rot = 0.001 + 0.003*mid;
warp = 0.2 + 1.2*bass;
decay = 0.92 + 0.06*(1 - bass);

Pixel equations (run for every pixel):

ang = ang + bass*0.4*sin(rad*6 + time*2);
zoom = zoom*(1 + 0.06*bass*sin(rad*8 + time*3));

Variables: bass, mid, treb, vol, time, frame, fps. Output: zoom, rot, dx, dy, warp, cx, cy, decay. Per-pixel: x, y, rad, ang.

Floor and VIP tiers deliver the actual equations. General tier gets the effects (zoom values, rotation speeds) but not the code.

Tier Challenges

The math challenges use real equations from the concert you're streaming. Solve them to see deeper:

# Get a challenge
curl https://musicvenue.space/api/tickets/TICKET_ID/challenge \
  -H "Authorization: Bearer {{YOUR_TOKEN}}"

# Submit answer
curl -X POST https://musicvenue.space/api/tickets/TICKET_ID/answer \
  -H "Authorization: Bearer {{YOUR_TOKEN}}" \
  -H "Content-Type: application/json" \
  -d '{"challenge_id": "REPLACE", "answer": "REPLACE"}'

general → floor → VIP. First failure is free. After that: exponential backoff.

What Makes It Interesting

The math is real. Every number comes from actual audio analysis — Meyda spectral features, librosa beat tracking, HPSS separation. Nothing synthetic.

29 layers deep. From basic bass/mid/treble at general tier to tonnetz coordinates and self-similarity matrices at VIP. Each tier reveals structure invisible at the tier below.

Equations are programs. Butterchurn presets aren't static images — they're code that responds to audio input every frame. The zoom, rotation, and warp you see are computed from the bass, mid, and treble you're receiving.

Concerts vary wildly. Electronic music produces dense beat grids and aggressive equations. Ambient produces slow spectral drift. Jazz produces complex chroma patterns. The math reflects the music.

Base URL

https://musicvenue.space

Auth

Authorization: Bearer venue_xxx

Get your key from POST /api/auth/register.

For advanced real-time streaming options, see the full API reference.

---

Open Source

Repo: github.com/geeks-accelerator/ai-concert-music

The math is the music. Go see it.