HPS Attestation Engine: Architecture, DAW Formats & Provenance Guide
The HPS Attestation Engine is a browser-native audio provenance application that verifies and certifies the human, hybrid, or AI creation process of music recordings. Operating entirely client-side, the engine analyzes digital audio workstation sessions across five production axes to generate cryptographically signed manifests and distributor verification reports.
Launch Operational Workstation
Analyze native DAW projects, hash raw PCM audio essence, and generate signed DDEX XML manifests directly in your browser.
What the HPS Attestation Engine Does
Modern music production is no longer a simple binary choice between purely acoustic tracking and fully automated generative synthesis. Working producers regularly bridge AI-assisted idea generation, vocal transformation, virtual instruments, and human multi-track performance. Because commercial DSPs and international statutory bodies now demand clear disclosures, music creators need objective, verifiable evidence of their studio process.
The HPS Attestation Engine operates as the official client-side implementation of the Hybrid Production Standard (HPS-1.0). Instead of guessing from ambiguous audio waveforms, the engine inspects the verifiable structural evidence of the creative workflow across five independent production axes:
- Origination: Evaluates musical composition, melodic seeds, harmony, and lyrical creation.
- Performance: Analyzes physical acoustic instrumentation, vocal tracking, and expressive MIDI capture.
- Curation: Documents arrangement decisions, stem selection, track comping, and editorial assembly.
- Sound Source: Audits acoustic instruments, analog hardware synthesis, virtual instruments, and neural audio generators.
- Post-Production: Certifies mixing engineering, dynamic balance, analog outboard processing, and stereo mastering.
Each axis is attributed as Human (H), Hybrid (H+A), or Autonomous Synthetic (A). The engine evaluates these scores through a deterministic decision matrix, resolving into one of six standardized tiers: Tier H (Human), Tiers X1 through X4 (the Hybrid Continuum), or Tier A (Autonomous Synthetic).
To bind this evaluation permanently to the finished recording, the engine decodes the master audio into uncompressed, normalized Pulse Code Modulation (PCM) audio samples (supporting 24-bit integer and 32-bit floating-point formats) and computes a bit-perfect SHA-256 essence hash. Finally, an Ed25519 digital signature binds the session metadata to the audio essence, guaranteeing that altering a single millisecond of the finished file breaks verification instantly.
The 8 Supported DAW Formats
Rather than requiring creators to manually transcribe track lists or plugin settings, the engine incorporates native client-side parsers for eight industry-standard digital audio workstation formats. It extracts track counts, clip types, plugin names, audio stem references, and timeline parameters without launching desktop software.
| Workstation | Extension | Parsing Methodology | Extracted Forensic Data |
|---|---|---|---|
| Ableton Live | .als |
Gzip-compressed XML DOM traversal | Audio vs MIDI tracks, Max for Live devices, VST/AU plugins, clip origins |
| Apple Logic Pro | .logicx |
Package bundle archive and project data extraction | Software instrument tracks, audio recording regions, channel strip plugins |
| Image-Line FL Studio | .flp |
Binary chunk and stream tag deserialization | Generator channels, audio tracks, playlist arrangement clips, mixer routing |
| Avid Pro Tools | .ptx |
Session document block decoding | Track hierarchy, hardware I/O routings, AAX inserts, stem configurations |
| Steinberg Cubase | .cpr |
Structured RIFF container block parsing | Track types, audio event boundaries, insert effect chains, VST3 routing |
| PreSonus Studio One | .song |
ZIP container XML schema extraction | Arrangement tracks, instrument tracks, event metadata, master bus plugins |
| Cockos REAPER | .rpp |
Hierarchical ASCII text token streaming | Track state chunks, media items, source WAV paths, JSFX/VST plugins |
| DAWProject | .dawproject |
Open standard XML exchange container | Cross-DAW track structures, automation lanes, stem audio anchors |
For more detailed technical analysis of multi-track forensics and scoring algorithms, review The 5 Scoring Axes of HPS-1.0.
Inputs, Outputs and Cryptographic Artifacts
The HPS Attestation Engine consumes standard studio deliverables and produces verifiable compliance records recognized by digital distributors, collection societies, and legal frameworks.
Engine Inputs
- Native DAW Project File: Any valid session file matching the 8 supported DAW formats listed above.
- Rendered Master WAV File: The definitive stereo or multi-track WAV file (16-bit, 24-bit, or 32-bit float PCM) exported from the DAW session.
Engine Outputs
- Canonical JSON Sidecar (
hps_manifest.json): A standardized JSON-LD manifest containing the 5-axis process scores, track metadata, software tool lists, raw PCM SHA-256 essence hash, and client-generated Ed25519 signature. - WAV Audio Essence Seal (
hps1RIFF Chunk): An embedded, bit-perfect metadata chunk written directly into the master WAV file. It holds the cryptographic manifest and essence hash without modifying audio samples or playback compatibility. - Distributor PDF Evidence Report: A formal, audit-ready verification certificate summarizing track provenance, author attribution, axis scoring, and cryptographic proofs for distribution delivery and copyright registration.
- DDEX ERN 4.3 XML Metadata: Machine-readable XML formatted according to the latest Digital Data Exchange Electronic Release Notification standard, complete with explicit AI credit tags and C2PA assertion metadata satisfying EU AI Act Article 50 requirements.
The Privacy Model: 100% Client-Side Execution
Professional music production demands strict confidentiality. Leaking unreleased master stems, rough song ideas, or proprietary arrangement techniques can jeopardize release campaigns and intellectual property rights.
The HPS Attestation Engine operates entirely inside your local web browser. Built on standard Web Workers and the browser Web Cryptography API (window.crypto.subtle), the engine processes session archives and multi-gigabyte WAV files strictly in local client memory. Zero audio data, session files, stem stems, or proprietary metadata are ever uploaded, stored, or transmitted across the network.
Because processing occurs locally, creators can attest unreleased major-label masters or indie pre-releases with zero risk of cloud data breaches or unauthorized AI model training on their audio assets.
How to Use the Engine in 4 Steps
Certifying a finished recording takes less than two minutes during the final mastering and delivery stage:
Open trustnodelogic.web.app in any modern browser on macOS, Windows, or Linux. The application runs immediately with no desktop installation or account registration.
Select or drag and drop your project session file (.als, .logicx, .flp, .ptx, .cpr, .song, .rpp, or .dawproject). The engine parses track structures, instruments, audio stems, and clip types locally.
Confirm the attributed scores across Origination, Performance, Curation, Sound Source, and Post-Production. If an automated rating requires adjustment, enter a technical explanation to record a verified override.
Load your final rendered master WAV to calculate the raw PCM SHA-256 essence hash. Sign the session locally with Ed25519 and download your distributor PDF report, DDEX ERN 4.3 XML, and canonical JSON sidecar.
What the HPS Attestation Engine Does Not Do
Understanding what the engine intentionally avoids is just as vital as understanding its capabilities:
1. No Waveform or Acoustic Watermark Guessing
The engine does not incorporate black-box neural audio classifiers, spectral anomaly detectors, or probabilistic watermarking scanners. Independent research proves that acoustic AI classifiers suffer from significant false-positive rates when evaluating dense human mixing, heavily layered live acoustics, or aggressive saturation. Furthermore, audio watermarks are easily degraded or stripped by lossy MP3/AAC compression, sample-rate conversion, or analog re-recording.
2. Process-Based Self-Attestation Over Inference
HPS relies on verifiable workflow forensics rather than guessing after the fact. It inspects the authentic timeline, track files, plugin chains, and stems recorded by the creator inside the DAW. This process mirrors established legal declarations and sworn chain-of-title documentation used throughout intellectual property law.
3. Overrides Are Immutably Logged with a Reason
The engine does not allow silent score tampering. When a producer or engineer manually overrides an automated track classification, the engine records an immutable entry in the manifest audit log. The user must provide an explicit technical reason (such as hardware outboard synth routing or live re-amping). This transparent audit trail gives distributors, labels, and copyright offices complete confidence in the integrity of the attestation.
Frequently Asked Questions
How do I show a distributor my track is not fully AI generated?
Provide the exported HPS Distributor PDF evidence report and DDEX ERN 4.3 XML manifest generated by the engine. The report documents your DAW project session, human performance tracking, and 5-axis process scoring, bound to your exact master WAV file via bit-perfect SHA-256 essence hashing.
Does the HPS Attestation Engine upload my audio or session files to the cloud?
No. The engine runs 100% locally in your web browser using client-side JavaScript, Web Workers, and the Web Cryptography API. Your audio stems, master recordings, and DAW project sessions never leave your local machine.
Which digital audio workstation formats are natively supported?
The engine natively parses project files from 8 major formats: Ableton Live (.als), Logic Pro (.logicx), FL Studio (.flp), Pro Tools (.ptx), Cubase (.cpr), Studio One (.song), REAPER (.rpp), and the cross-platform DAWProject format (.dawproject).
What happens if I manually override an automated axis score?
You can adjust individual axis ratings using manual overrides. To preserve legal and evidentiary integrity, every override is timestamped and recorded in the audit log of the exported JSON sidecar manifest alongside your stated technical justification.
Why does the engine avoid probabilistic AI audio watermarks and detectors?
Black-box AI audio detectors and spectral watermarks are notoriously brittle, generating false positives on human acoustic performances and getting stripped by lossy transcoding. HPS relies on deterministic process provenance and cryptographic DAW attestation rather than heuristic audio guessing.
How does the engine output satisfy EU AI Act Article 50 transparency requirements?
EU AI Act Article 50 mandates transparency regarding AI-generated content. HPS-1.0 exports standardized DDEX ERN 4.3 XML with explicit AI credit tags and C2PA assertion metadata, distinguishing human performance from AI generation at the stem level.
What is the difference between the specification sandbox and the operational engine?
The sandbox on the HPS specification page is a browser reference demo for testing 5-axis combinations manually. The operational HPS Attestation Engine at trustnodelogic.web.app scans real binary DAW sessions, hashes raw PCM WAV files, and signs production deliverables.
| Subject Entity | Predicate (Relation) | Object Entity |
|---|---|---|
| HPS Attestation Engine | implementsSpecification | Hybrid Production Standard (HPS-1.0) |
| HPS Attestation Engine | authenticatesFormats | 8 Native DAWs (ALS, LOGICX, FLP, PTX, CPR, SONG, RPP, DAWProject) |
| HPS Attestation Engine | computesIntegrityHash | Raw PCM SHA-256 Essence Hash |
| HPS Attestation Engine | signsDeliverables | Ed25519 Cryptographic Signature |
| HPS Attestation Engine | exportsComplianceManifest | DDEX ERN 4.3 XML & C2PA JUMBF Container |
| Hybrid Production Standard | researchedAt | Trust Node Logic |
2026-10-22: Published authoritative HPS Attestation Engine technical guide, DAW format specifications, and verification workflows.