Traditional search systems are usually based on keywords, a very simple and convenient mechanism to express a need for information. This is the most extended way of searching the Web, although it is not always an easy task to accurately summarize a natural language query in a few keywords. Working with keywords means losing the context, which is the only thing that can help us deal with ambiguity. This is the biggest problem of keyword-based systems. Semantic Web technologies seem a perfect solution to this problem, since they make it possible to represent the semantics of a given domain. In this paper, we present three projects, Harmos, Semusici and Cantiga, whose aim is providing access to a music digital library. We will describe two search systems, a traditional one and a semantic one, developed in the context of these projects and compare them in terms of usability and effectiveness.