Anubis

Audio watermarking that survives real-world attacks.

Track leaks even after pitch shifts, compression, re-uploads, and edits — without affecting audio quality.

Built for producers, labels, and distributors who need to trace leaks without changing how they work.

Protect your tracks. Trace leaks. Verify sources.

Scroll

Transparent by design. Proven on challenging, production-quality audio.

These results are from testing on a difficult, sparse track — the kind of audio where watermarking artifacts are easiest to hear and hardest to hide.

+0.04
ODG (Perceptual Quality)
Transparent
4.73/5
ViSQOL MOS-LQO
Transparent
~38 dB
SNR
-27 dB
NMR
Well masked
Even on minimal, exposed audio, the watermark remains inaudible.

Tested on real tracks — from sparse acoustic recordings to heavily processed electronic audio.

Original vs Watermarked — can you tell the difference?

Toggle between original and watermarked versions. The watermark is embedded across the full spectrum — yet remains inaudible under critical listening.

Sparse Acoustic

Minimal production — where artifacts are easiest to hear

Playing: Original

Hip-Hop

Mastered commercial track with vocals and bass

Playing: Original

Electronic / EDM

Dense, heavily processed production

Playing: Original

Audio files coming soon — add your own .wav files to the /public/audio/ directory.

Validated across a range of audio styles

Dense Electronic / DnB
Hip-hop & Mastered Commercial
Acoustic & Vocal-heavy
Sparse, Minimal Productions
Consistent transparency and decode accuracy across all tested material.

Built to survive what actually happens to audio

Not just lab conditions.

Tested Against

Trimming (milliseconds to multiple seconds)
Time shifting / alignment changes
Compression (MP3, AAC, Opus)
Resampling (44.1kHz / 32kHz round-trips)
EQ (lowpass / highpass filtering)
Dynamic processing (compression, limiting)
Gain changes & clipping
Noise injection

🔁Advanced Transform Resistance

Pitch shifting
up to ±6 semitones
Time stretching
up to ±15%
Stacked attacks
multiple transformations combined
Piracy workflow survival
Adobe Premiere → YouTube upload → YouTube-to-MP3, SoundCloud rips, and similar real-world piracy chains
Designed to handle actual edits — not just simple transformations.

Extensively validated via PEAQ, ViSQOL & A/B/X testing

Anubis has been rigorously validated using industry-standard objective metrics (PEAQ, ViSQOL) alongside controlled A/B/X listening tests. Below is a sample pool drawn from a random selection of files spanning various genres, production styles, and quality levels.

35
Sample files tested
30
Perfect decodes
≤2
Near-perfect bit errors
100%
Identified across tested cases
Even under heavy distortion, Anubis still recovers the correct identity.

Resilient across industry-standard formats

MP3 (320k → 80k)
Stable
AAC / Opus
Tested & stable
Multi-pass codec chains
Verified
Extreme low bitrate (64k)
⚠️Degraded but recoverable

More than decoding — Anubis understands degraded audio

Error Correction (ECC)

Restores damaged payloads

Multi-segment Consensus

Consensus decoding across segments

Confidence Scoring

Based on statistical agreement

Database Matching

Identifies the most likely source

Even when audio is heavily altered, Anubis can still tell you where it came from.

Trace leaked tracks back to specific recipients

Protect unreleased music and promos

Verify ownership without needing original files

Identify sources even after editing, compression, or re-encoding

Three steps. That's it.

1

Encode

Upload your track and assign it to a recipient

2

Distribute

Send your audio as normal

3

Decode

Upload any version to instantly identify the source

Completely blind detection — no original file needed.

Unlike many watermarking systems, Anubis operates completely blind — only a clip of the watermarked track is needed to decode. A longer, complete watermarked track will yield a stronger decode, but Anubis can operate on clips as short as 15 seconds.

Anubis was built with a "producer first" philosophy. Before anything else, a key requirement was that the watermark must remain inaudible under critical listening. Using novel approaches, we've achieved impressive transparency results across all types of content — from sparse acoustic recordings to dense, mastered productions.

Embed identity into your audio — and prove it, no matter what happens to the file.

Anubis is launching soon — get in early.