Overview

This page is designed to contain additional resources related to a conference paper submitted to IEEE MWSCAS 2000 entitled, "Performance Evaluation of Digital Audio Watermarking Algorithms."  Copies of the paper are available in Postscript and PDF formats.

Audio Samples

Below are ten samples of wideband audio signals that we used in our research.  Each signal is a raw WAV file of ten seconds in length, sampled at 44.1 kHz and quantized to 16 bits per sample (CD quality, in other words).  Each file is rather large, over 800 kb, so please be patient with our web server.

The samples were selected to represent different types of popular music in an attempt to determine how well watermarking algorithms work across a range of signals with different spectral characteristics.

• Classical 1
• Classical 2
• Country 1
• Country 2
• Folk 1
• Folk 2
• Pop 1
• Pop 2
• Blues 1
• Blues 2

Watermarked Samples

As described in our paper, we reviewed five algorithms for embedding data within audio sequences.  Below are watermarked versions of some of the samples using the algorithms.

Echo Coding

• Blues 1

Phase Coding

• Classical 1

Direct-Sequence Spread Spectrum

• Country 1

Frequency-Hopped Spread Spectrum

• Pop 1

Frequency Masking

• Folk 1