Skip to content

Watermark Me!

Ever since Steve Jobs published his Thoughts on Music the music industry and its DRM partners have been in a tizzy, and it’s clear why that is. But there’s a solution, if only we can save them from themselves.

The biggest problem with DRM is lock-in to a particular device or class of devices. Getting one’s music to another type of device is impossible without a DRM break or incurring quality loss. But that’s not what the *AA says is the purpose of DRM – they say it’s to prevent copying. They’re lying.

Right now if you do manage a DRM break of any of the existing systems (and they all fall eventually) you get a media file that’s identical to the file on anybody else’s computer if they do the same thing. From there one can put the file up on a p2p network, the boogeyman of the argument. It’s impossible to trace the origin of the DRM-stripped file, which makes prosecuting copyright law difficult and makes companies like BayTSP a small mint.

Why are these p2p networks so popular? Part of it is the close-to-zero cost, sure, but many others participate in the networks to get content online a format they can manage (typically MP3). AllOfMP3.com (Mothra to p2p’s Godzilla) is popular for similar reasons. No matter how many times I tell these people to ‘just buy the CD’, I can’t stop the march of the world online, nor should I expect to.

So, the obvious solution is to make each media file unique, and trackable. The tool to do this comes from the field of steganography and is called Audio Watermarking. The theory goes like this: an amount of data is hidden in another amount of data, in this case the digital data of an audio (or video) file. It’s a field of active research and the results are good. There are concerns about the quality effect hiding extra data in something like an audio file can have, and you can see this with lossless codecs and an oscilloscope, but for Internet-based audio, where the files are compressed, it has been shown that the data hiding induces less change in the resulting waveform than the compression does, so you effectively can’t hear the watermark. A good watermarking algorithm is resistant to modification, sample rate conversion, etc, and it’s known how to do this.

So, what to include in the watermark? Well, you can include a user ID from the store where it was purchased. You can include my name and home phone number for all I care, my files are only going to be on my devices. The trick here, I believe, is to work out a fair system where the encoding specifies the privacy demands of the end user, such that players can’t become tracking devices without user consent. I suspect the W3C’s P3p can be repurposed for this task. These preferences would need to be embedded in the media file itself. To make all of this work, a standards organization (like the MPEG or ISO) needs to define a common standard container/structure format.

The place where this information is encoded is a classic engineering problem – if it’s done on the server that’s more secure. If it’s done on the client end that’s cheaper for the vendors but somewhat less secure.

So, what do we wind up with? A system for purchasing media on the the Internet that creates files that are playable on all devices, can be easily moved/converted/archived, can be edited and sampled for derivative works and other Fair Use applications, will expire gracefully from Copyright protection at the end of their Copyright term, and allow for prosecution of violations of Copyright Law. This gives something to everbody: the users of the media have it much easier and don’t have to worry about lock-in or obsolescence problems, therefore they buy more music and the vendors (e.g. iTunes) sell more music and devices (e.g. iPods), therefore the *AA get more royalties, and even companies like Macrovision can take their shot at peddling encoders.

So, what’s not to like? I’ll tell you – the *AA thinks DRM will force us to re-purchase our favorite media files over and over each time a new generation of devices comes out. That’s a powerful high they’re on, but users of their content will only purchase Back To The Future so many times before they get as mad as hell. Once these cartels get some rehab about this (and it might take a legislature, unfortunately, to do the rehab) audio watermarking is going to be waiting to make this market explode.