Publishing data to the web in RDF is explained in this tutorial:
For small scale data, publishing need be no more than translating your data into RDF in a single file, and putting that RDF file on your web server.
For larger scale data, you may want to put all of it within a SPARQL end point. If your data is currently in a relational database, you can set up a SPARL end point automatically with D2R. A SPARQL end point has the advantage of being able to support more complex queries: see http://dbtune.org/jamendo/ for an example.
Typically you will want to link your data with existing datasets - for example to link your specific data about particular recordings to high-quality standard metadata, so that anyone using your data can be confident that they know exactly which tracks it relates to. omras2 offers a tool, GNAT, that automates this particular process of linking audio tracks to metadata in the MusicBrainz catalogue. Note that you can publish data linked to MusicBrainz URIs in this way without having to publish the audio itself.
Other interesting links might include linking composers of classical pieces to their biographical entries in DBpedia.