The MUSART Testbed for Query-By-Humming Evaluation

Abstract

Evaluating music information retrieval systems is acknowledged to be a difficult problem. We have created a database and a software testbed for the systematic evaluation of various query-by-humming (QBH) search systems. As might be expected, different queries and different databases lead to wide variations in observed search precision. "Natural" queries from two sources led to significantly lower performance than that typically reported in the QBH literature. These results point out the importance of careful measurement and objective comparisons to study retrieval algorithms. We compare string-matching, contour-matching, and hidden Markov model search algorithms in this study. An examination of scaling trends is encouraging: precision falls off very slowly as the database size increases. This trend is simple to compute and could be useful to predict performance on larger databases.