AIAPS-001 — Core Specification
Active StandardAIAPS Specification
This document defines AIAPS — a standard for protecting human-made audio recordings from unauthorized AI training and synthetic reuse. It specifies the requirements for fingerprinting, signature embedding, registration, and verification.
Section 01 — Terms and Definitions
Definitions
AIAPS Fingerprint
A unique identifier derived from the spectral and temporal structure of a recording. The fingerprint is a compact numerical representation that enables precise identification and matching.
AIAPS Signature
A hidden authenticity marker embedded directly into the audio signal. The signature is song-specific, derived from the recording itself, and embedded in psychoacoustically masked regions to ensure inaudibility.
AIAPS Registry
The authoritative registry of AIAPS-conformant recordings. The registry stores fingerprint hashes, timestamps, verification identifiers, and usage policy declarations.
AIAPS Conformant
A human-made recording is AIAPS-conformant when it has been processed according to this specification, contains a valid embedded signature, and is registered in the AIAPS registry.
Verification
The process of analyzing a recording to detect an AIAPS signature, matching its fingerprint against the registry, and confirming its conformance status and usage policy.
Usage Policy
A machine-readable and human-readable declaration attached to an AIAPS-conformant recording that specifies the terms under which the recording may or may not be used, particularly with respect to AI training and synthetic generation.
Section 02 — Normative Requirements
Core Components
An AIAPS-conformant implementation SHALL include all of the following components to protect human-made recordings.
2.1
Fingerprint Generation
The implementation SHALL analyze the spectral and temporal structure of the input recording and generate a unique perceptual fingerprint. The fingerprint SHALL be deterministic: the same recording SHALL always produce the same fingerprint.
2.2
Signature Embedding
The implementation SHALL embed a song-specific authenticity signature into the audio signal. The signature SHALL be placed in psychoacoustically masked regions to ensure inaudibility. The signature layout SHALL be derived from the structure of the recording itself.
2.3
Registry Registration
The implementation SHALL register the recording in the AIAPS registry. The registry entry SHALL include: a fingerprint hash, a registration timestamp, a unique verification identifier, and a usage policy declaration.
2.4
Verification
The implementation SHALL provide a means to verify any recording against the AIAPS registry. Verification SHALL include signature detection, fingerprint matching, and registry confirmation.
2.5
Usage Policy Attachment
Every AIAPS-conformant recording SHALL carry a usage policy notice. The default policy SHALL declare that the recording is not authorized for AI training, dataset ingestion, voice or style cloning, or synthetic generation without explicit permission from the rights holder.
Section 03 — Usage Policy
Default Usage Restriction
The following notice SHALL be attached to all AIAPS-conformant recordings.
AIAPS Protection Certificate
VerifiedAIAPS Protected — No AI Training
Usage Policy
This recording is registered with AIAPS and may not be used for AI training, dataset ingestion, voice or style cloning, or synthetic generation without permission from the rights holder.
AIAPS Registry — Authentic Record
Rev. 1.0
The policy notice SHALL appear in:
01Registry records
02Audio file metadata
03Creator certificates
04Verification results
Section 04 — Architecture