Heart-rate waveform morphing into an audio waveform

Google Gemini XPRIZEAugust 2026Confidential

Physiological fingerprinting for AI-generated music

The rights management engine that connects YouTube Content ID, training-data attribution, and human biology — one auditable pipeline from heartbeat to royalty.

Try the live demo See the pipeline $2.5M seed round

Every beat: is a prescription
6-phase: Gemini prescription
<2s: prescription latency
100%: AI disclosure

AeroGlyphics — VO₂max music videoShowcase clip — audio prescribed from a VO₂max / heart-rate zone inputSource

Presented by Eangelica Aton · AeroGlyphics / GoApercu · Stanford-trained

The problem

Three converging challenges for AI-generated music

Rightsholders can now trace influence, platforms enforce automatically, and regulators require disclosure. Nothing connects those three pressures to the person the music was made for.

Rightsholder liability

Sony's AI music detection can trace which training data influenced an output — e.g. "30% Beatles, 10% Queen".

Enforceable copyright claims

Platform enforcement

YouTube Content ID claims against AI-generated music are up 40% since 2024; Spotify rejects unlabeled AI tracks.

Loss of distribution channels

Regulatory compliance

The FTC AI disclosure mandate requires visible "AI-generated" labeling at the point of consumption.

Legal exposure and fines

The gap

No system connects creative provenance — "generated from this user's VO₂max and heart rate zone" — to training data provenance — "influenced by these copyrighted works." That makes license negotiation and royalty distribution impossible at scale.

The solution

A verifiable link between human physiology and generated audio

Embed immutable, auditable metadata that ties a user's physiology to the generated audio — then register the asset with Content ID and, optionally, run it through an attribution engine.

User biometrics

VO₂max
Heart rate
HRV
Mood
Age

Gemini prescription

6-phase arc
BPM targets
Zone sequence
Science rationale

AI music generation

Suno API / custom model
30–90 second track
Phase-matched

Rights management

Metadata injection
Digital watermark
Audio fingerprint
YouTube Content ID

Audit trail

Immutable PostgreSQL log · prescription ID → audio mapping · FTC disclosure status.

Attribution engine

Sony · Sureel · Musical AI — estimates training data influence and enables license negotiation.

How it works

Step 1 — Generation and metadata

The user completes a session, Gemini 2.0 Flash returns a six-phase prescription, and the music engine renders a 30–90 second track. Then the fingerprint goes in.

Metadata field	Example value	Purpose
physiological_prescription_id	550e8400-e29b-41d4-a716-446655440000	Links to session audit trail
vo2max_quintile	High	Base prescription intensity
target_hr_zone	4	Zone classification
mood	Push Me	User intent
is_ai_generated	true	FTC compliance
generation_platform	VO₂max.One	Platform attribution
generated_timestamp	2026-08-14T14:30:00Z	Time of generation

FFmpeg — ID3 metadata injection

ffmpeg -i generated_audio.wav \
  -metadata artist="VO2max.One" \
  -metadata title="Push Me Session" \
  -metadata prescription_id="550e8400-..." \
  -metadata vo2max_quintile="High" \
  -metadata is_ai_generated="true" \
  -codec copy output_with_metadata.mp3

Digital watermark

Imperceptible audio watermark, 3GPP standards-aligned.
Survives transcoding, compression, and analog-to-digital conversion.
Carries the same physiological_prescription_id as the headers.

Defense in depthMetadata-strip resilient

Step 2

Fingerprinting and rights registration

A probabilistic excerpt-based fingerprint — the industry standard, compatible with YouTube Content ID — identifies the track even when metadata is stripped. Registration requires YouTube Partner Program access.

Step	Action	API endpoint
1	Create/update content asset	`POST /youtube/v3/assets`
2	Upload content reference	`POST /youtube/v3/contentReferences`
3	Set ownership / policy	`POST /youtube/v3/assetOwnership`
4	Map asset to user prescription	`store asset_id in audit trail`

Policy configuration

Monetize — earn from ad-supported views
Block — prevent unauthorized uploads
Track — monitor usage without enforcement

Asset metadata

VO₂max.One — Push Me (High Quintile). AI-generated adaptive music for Zone 4–5 cardiovascular training.

#AIGenerated#VO2max#Fitness#PrescriptionMusic

Step 3

Attribution and royalties

After generation, the track can pass through a third-party attribution engine that estimates which copyrighted training data influenced the output — the evidence base for license negotiation.

Attribution engine	Focus area	Availability
Sony AI Music Detection	Music style attribution (R&D)	Announced Feb 2026
Sureel	Artist style detection	Available · partnered with STIM
Musical AI	Music attribution	Commercially available
Neutune	Music attribution	Available (South Korea)

Attribution output

{
  "prescription_id": "550e8400-...",
  "attribution_results": [
    { "artist": "The Beatles", "influence_percent": 30 },
    { "artist": "Queen", "influence_percent": 10 },
    { "artist": "Daft Punk", "influence_percent": 15 },
    { "artist": "Unknown/Other", "influence_percent": 45 }
  ],
  "attribution_engine": "Sony_AI_v2",
  "attributed_timestamp": "2026-08-14T14:35:00Z"
}

For B2B partners

Transparent royalty reporting to rightsholders.

For license negotiation

Auditable evidence for fair use claims.

For investor confidence

Proactive rights management, built in.

Compliance

Meeting FTC, platform, and rightsholder requirements

Disclosure is not a footer note. It is surfaced at the point of consumption and written into every distribution channel's metadata.

FTC disclosure mandate

Clear, visible labeling of AI-generated content at the point of consumption.

In-app: "This music was generated by AI for your personalized workout," shown during playback.
YouTube: "AI-generated" tag auto-injected into the description.
Spotify: ai_generated flag set at submission.
TikTok: dedicated AI music label via API.

Platform	Requirement	Implementation
YouTube	Disclosure in video description	Auto-inject tag via API
Spotify	`ai_generated`	Set flag during submission
TikTok	Dedicated AI music label	Apply label via TikTok API

Immutable audit trail — PostgreSQL

Column	Type	Description
prescription_id	UUID	Primary key
user_id	UUID	Anonymous user ID (or identified for B2B)
vo2max_value	FLOAT	VO₂max estimate (ml/kg/min)
vo2max_quintile	VARCHAR	Low / Moderate / High / Elite
target_hr_zone	INT	1–5
mood	VARCHAR	push / dance / run / move / challenge
gemini_prescription_json	JSONB	Full Gemini output
generated_audio_url	VARCHAR	S3 / CDN URL of generated audio
youtube_content_id_asset	VARCHAR	YouTube asset ID
attribution_results	JSONB	Attribution engine output (if available)
is_ai_generated_disclosed	BOOLEAN	FTC compliance flag
generated_timestamp	TIMESTAMP	ISO 8601

Market positioning

Why this creates a defensible moat

VO₂max.One is the only product connecting human physiology to AI rights management. Not Spotify Running, not Apple Fitness+, not any fitness app has integrated physiological metadata, Content ID registration, and attribution tracking.

Moat	Description	Competitive impact
Audit trail	Immutable log of every track's physiological inputs and outputs	Enables license negotiation; compliance moat
Content ID integration	Ownership established at the time of generation	Prevents unauthorized use; monetizes distribution
Attribution integration	Optional training data contribution estimation	Transparent royalty distribution
Regulatory compliance	FTC disclosure and platform ToS from day one	Reduces legal risk; responsible partner

Revenue model synergy

Revenue stream	Rights management role
Consumer subscription — $12.99/mo	Direct-to-consumer; visible AI disclosure in UI
B2B wellness — $5/employee/mo	Enterprise audit trail for compliance reporting
API / data licensing — $0.50–$2/user/yr	Attribution results for research
Content monetization — YouTube ads	Asset monetization via Content ID policy

Technical architecture

System components and data flow

Four layers, one identifier. The prescription ID created at generation follows the audio through fingerprinting, registration, attribution, and audit.

User layer

iOS / Android app (Expo)
Apple HealthKit
Fitbit Cardio Fitness Score Web API

Prescription & music generation

Gemini 2.0 Flash · temperature 0.3 · structured JSON
6 phases: warm-up → zone 2 → threshold → recovery → peak → cool-down
AI music engine · Suno API / custom model · 30–90s

Rights management & fingerprinting

Metadata injection (FFmpeg)
Digital watermark (3GPP)
Audio fingerprint (excerpt)
YouTube Content ID asset registration
Attribution engine (optional)

Audit & compliance

PostgreSQL immutable prescription audit trail
prescription_id → audio mapping
FTC disclosure status · Content ID asset ID

Roadmap

Twelve months, four phases

From metadata injection to a fully automated rights management pipeline and B2B API.

Q1 2027

Foundation

Metadata injection MVP (2 weeks)
Digital watermarking POC (3 weeks)
Audit trail database (2 weeks)
FTC disclosure UI (1 week)
Legal review (6 weeks pre-launch)

Q2 2027

Platform integration

YouTube Partner Program access (4–8 weeks)
Content ID API integration (4 weeks)
Spotify metadata compliance (2 weeks)
Attribution engine POC (6 weeks)
Partner pilot (4 weeks)

Q3 2027

Scale & automation

Automated fingerprinting pipeline (4 weeks)
Attribution engine full integration (6 weeks)
Policy automation (3 weeks)
Compliance audit (ongoing)

Q4 2027

Advanced features

"Virtual DJ" prescription playlist (6 weeks)
Offline pre-generation (4 weeks)
B2B API for partners (8 weeks)
Series A readiness: 100K users, $1M ARR

Key milestones

Milestone	Timeline	Success metric
Metadata injection MVP	2 weeks	Track generation with embedded physiological metadata
YouTube Partner Program access	4–8 weeks	Approved partner; Content ID API access
Content ID API integration	4 weeks	Asset registration and policy management
Attribution engine POC	6 weeks	Attribution results stored in audit trail
Series A readiness	Q4 2027	100K users, $1M ARR, automated rights management

Risks & mitigations

Managing technical and regulatory challenges

Risk	Likelihood	Impact	Mitigation
YouTube Partner Program access denied	Medium	High	Apply early; partner with an existing YPP member
Attribution engine accuracy	High	Medium	Position as "best-effort estimation"; transparent methodology
Disclosure requirements change	Medium	High	Legal counsel 6 weeks pre-launch; quarterly reviews; flexible UI
Rightsholders challenge attribution	Medium	High	Immutable audit trail; multiple engines; negotiation over enforcement
Platform ToS change	Low	High	B2B partnerships as primary channel; keep direct-to-consumer app
Digital watermark cracked	Low	Medium	Open standards; watermarking plus robust fingerprinting

The ask

$2.5M seed round — August 2026

To build the first physiological fingerprint and rights management engine for AI-generated music — connecting user biology, platform compliance, and rightsholder licensing in a single auditable pipeline.

40%

Engineering

Fingerprinting pipeline, Content ID, attribution, B2B API

25%

Partnerships & compliance

YPP application, legal, licensing negotiation

20%

Growth & distribution

App Store launch, content marketing, B2B partners

15%

Operations

Infrastructure, legal and regulatory compliance

Phase	Users	Run rate	Key achievement
Q1 2027	1K	$10K MRR	Metadata injection and audit trail live
Q2 2027	5K	$45K MRR	YouTube Content ID integration; partner pilot
Q3 2027	25K	$200K MRR	Automated fingerprinting; full attribution
Q4 2027	100K	$1M ARR	Series A ready; B2B API live