audiowmark - Audio Watermarking

About

audiowmark is an Open Source (GPL) solution for audio watermarking.

Here is a quick list of the main features:

blind decoding
watermark detection does not need the original file
128-bit message
the watermark can store 128 message bits
inaudible
informal listening tests were conducted, almost all users do not percieve the difference between original and watermarked signal
robust
we use forward error correction to ensure that a damaged watermark can still be decoded, so the watermark usually survives lossy compression (mp3/ogg)
secure
the algorithm is known, but a key ensures that others cannot read/modify watermarks generated by you with your private key
streaming
when delivering content to the user, we can watermark the audio on-the-fly
fast
audiowmark is implemented in C++, which makes adding and retrieving watermarks fast
open source
the code is provided under the GNU Public License version 3 or later

Documentation

This README contains the main documentation. There is also Documentation for Developers which describes how audiowmark works internally.

Source Code

Stable release:

Git repository access:

Discussion Group

Demos

Here are few watermarked demo songs, so you can compare the original files with the watermarked files. From our informal listening tests, we believe that almost all users do not percieve the difference between original and watermarked signal.

ArtistSongGenreOriginalWatermarked
STEEP It Could Be You Rock mp3 mp3
LukHash Take Control Electronic mp3 mp3
Brunela Crochenci Háblame - (Talk to me) Jazz mp3 mp3
The Riders Bad & Evil Metal mp3 mp3

The music used for these demos is from Jamendo and is under Creative-Commons License (Attribution/Non-Derivative/Non-Commercial).