ASPIRE LabIndiana University Bloomington Start with your task

OpenAudioMark · from research to readiness

Choose an audio watermark.
Deploy with confidence.

Start the guided path Explore the benchmark

Turn fragmented audio watermarking research into a practical path: define the job, compare observed evidence, and leave with a deployment plan.

10 methods 3 attack families

Comparable evidence · interactive listening lab

Test watermark robustness

Digital demo

Digital distortion

Loading the digital audio collection…

Loading the digital audio collection…

Each dataset follows one source through clean, watermarked, and distorted stages.

Why the guided path matters

No evaluated method survives everything.

51.63%

AI-based defenses

Share of evaluated attack settings defended by AI-based methods in aggregate.

28.97%

Traditional defenses

Share defended by traditional signal-processing approaches in aggregate.

2

Persistent weak points

Pitch shift and AI regeneration remain major failure modes.

Open-source methods

Ten different ways to leave a watermark in your audio.

Neural encoders and classical signal processing use different frequency coverage, message repetition, and training strategies.

Choose any method above to load its source-matched listening comparison in the demo.

Interactive Watermarking Demo · Docker + Gradio

Move from benchmark to prototype.

Run the complete testbed locally to upload your own audio, embed one or more watermarks, apply digital attacks, and validate the first version of your deployment plan.

Quick start 
./install.sh
Get deployment code View setup instructions
  1. 1

    Get the code

    Open the deployment repository above and download or clone the complete testbed.

  2. 2

    Start Docker

    Launch Docker Desktop, then run ./install.sh from the downloaded project folder. The first build may take 20–40 minutes.

  3. 3

    Open the Gradio interface

    When setup finishes, visit http://localhost:7860, upload your audio, and run the full evaluation.