Notated musical scores provide a basic framework on which a performance is realized. Hidden behind this fixed score is a dynamic unwritten set of performance rules and conventions which performers draw upon when generating their own personal interpretation of the musical work. With the advent of recording, studying this implicit aural tradition is made possible through direct analysis of recordings. In this work, we exploit the availability of numerous recordings of the same piece and propose a lightweight, unsupervised and audio-based approach focusing on the analysis of expressive timing. In particular, we align multiple performances of the same piece by different performers and study timing deviations between aligned performances. This way, we gain insight into the degree of individualism in expressive musical timing displayed throughout a piece, which is part of the unwritten musical domain which performers add to the music. A qualitative study of the results of our technique applied to five Chopin mazurkas shows that timing individualism can be related to musical structure at both higher and lower structural levels, and highlights interpretation aspects that cannot objectively be found from a musical score only.
Liem, Cynthia; Hanjalic, Alan; Sapp, Craig
Affiliations: Delft University of Technology, Deft, The Netherlands; Stanford University,Stanford, CA, USA(See document for exact affiliation information.)
AES Conference: 42nd International Conference: Semantic Audio (July 2011)
Paper Number: 6-1
Publication Date: July 22, 2011
Subject: Music Information Retrieval
No AES members have commented on this paper yet.
If you are not yet an AES member and have something important to say about this paper then we urge you to join the AES today and make your voice heard. You can join online today by clicking here.