By the use of suitable data-capture methods such as optical music recognition for
printed 16c part-books, combined with manual encoding for manuscript and other
sources, we can study large amounts of score-like musical data using computational
techniques for pattern-matching, classification and analysis. In this paper we look at the
ways such search methods can be applied across a mixed corpus of encodings of 16c
vocal and instrumental music, and how they might be extended to include audio
recordings. Although in principle an armoury of pattern-matching algorithms has
become available through decades of work in bioinformatics and information retrieval,
each one of these needs to be carefully reconsidered and possibly adapted for use in a
musicological context. Using a geometric pattern-matching method designed with music
in mind, we carried out a preliminary study of the ways in which ornamental patterns or
clichés were applied in the process of arranging pieces of vocal music for the favourite
16c instrument, the lute, showing that this was done differently for sacred and secular
music. Searching directly for simple melodic fragments within lute pieces is perhaps
unexpectedly challenging, mainly because of the number of ‘false positives’ one is likely
to encounter in any passages of dense polyphony. This forces us to adopt
computationally expensive strategies such as n-gram searching, with concomitant issues
around the analysis and display of results in a manner which would lead a musicologist
to a valid intuitive response. Further work is likely to involve examination of the internal
details of instrumental music notated in tablature, the system used by the lute, which
allows estimation of the relative difficulty of passages, or to ‘grade’ whole pieces. A
nearly identical tablature system is used today in vast quantities on the internet to share
interpretations of popular music for guitar and other instruments