Keeping collections of audio files in a net-accessible way is obviously
not a new idea. The simplest way to publish one's music collection to the
net is just to make it accessible as a WWW or FTP archive. Many people
do this today, but the utility of unconnected collections of audio is low.
In order to make these collections more useful, dedicated MP3 search
engines such as mp3.lycos.com have appeared. These search engines
try to be your- ``one-stop shopping'' for MP3's, by telling you where on the
Internet you can find your favorite pirated songs.
More recently, commercial jukebox products have become available that allow
you to organize and play locally-stored MP3's, but these products typically
do not permit sharing between users,
nor do they offer collaborative or interactive features.
This simplest kind of jukebox system is missing a number of benefits that the Ninja Jukebox provides. Simple directories of MP3 files offer no cohesive framework for security-related features such as authenticated or pay-per-use access. In addition, our component architecture allows the SoundSmiths to be active and easily updatable participants in the transmission of the audio, as opposed to merely serving a static file. This allows features such as transparent format conversion (.wav files on file systems, raw audio on CD-ROM drives, or MP3 files on file systems) and support for multiple transport mechanisms (streamed audio over HTTP, or VAT audio over a multicast IP channel).
SHOUTcast Another approach has recently come from [11]. is an ``Internet radio'' system that allows a site to serve an audio stream that can be picked up by multiple clients. The clients have to listen to what the servers decide to play; they have no way to interact with the servers. Although each server offers the same real-time audio stream to each of its clients, and though its name would imply something more clever, its underlying technology is just multiple simultaneous unicasts of the same data. servers communicate with one or more central databases in order to register the names of the programs they are currently ``broadcasting''. These databases can be queried by client programs (like MP3Spy [13]) to allow users to choose what channels they would like to hear.
The largest difference between and our work is that our goal was to provide a communal, collaborative, interactive jukebox, as opposed to a passive Internet radio station. That having been said, however, it would be possible for a SoundSmith to transmit any particular song over a true multicast channel [7]. servers also do no user authentication; one might indeed imagine that an Internet radio service would have no need for such a thing. However, given the broad view of ``authentication'' taken by the Ninja Jukebox, one could see that implementing, for example, subscription-based access or pay-per-use access, could add value (better quality of service, for example) even to a non-interactive service like Internet radio.
A related approach is the Interactive Multimedia Jukebox [1, 2], a system that allows one to add a measure of interactive preference feedback to traditional broadcast paradigms.
More recently, the SDMI project is starting to tackle the issues associated with copyright control and rights management, using a combination of tamperproof hardware (or software!) on the client end as well as watermarking and other technologies [14]. We view SDMI as largely orthogonal to our work: we have focused on building a music delivery service, rather than on what is done after the music has been delivered.
There have been a number of projects involved in the delivery of audio and/or video over digital networks (for example, [5]); these projects mainly concern themselves with the technology of media delivery. In contrast, we have left that issue largely unaddressed, as it is orthogonal to our own goals; we were more interested in the mechanisms of the service, rather than the mechanisms of serving.