Multimodal cross-linking

Cross-linking musical resources is relevant when aiming for search and browsing across collections and modalities. This can be considered in several forms. First of all, cross-linking can be performed between musical information in multiple modalities, considering the same piece. This links to existing work in the field of audio-to-score alignment (Arzt et al., 2015).

Secondly, cross-linking can be performed between musical information and related information or query descriptions in other modalities, which may not consider the same musical object anymore. Proper links may not only occur at the musical document level, but also at the fragment level. This problem is currently researched in the fields of music and multimedia hyperlinking (Eskevich et al., 2013, Sutcliffe et al., 2016).

Finally, current musical repositories are very much focused on musical information only, while in order to understand the societal context and meaning of music, it is relevant to also connect these resources to contextual non-musical resources, such as usage in soundtracks (Liem, 2016) and public expressions of listening and liking (İren et al., 2016).

TROMPA will realise further linking between musical and non-musical resources, on all three aspects of cross-linking. Regarding the alignment of symbolic information to audio, we will particularly research human-in-the-loop ways for giving feedback to alignment quality, forming an HCI and crowd-oriented complement to existing algorithms for alignment. Regarding the linking to non-musical resources, we will depart from potential connections to other professional collections through CDR’s structured metadata repositories, together with automatically generated consistent multilingual metadata specifications (Dingjan et al., 2017).  Regarding societal context and meaning, we will build on our existing work using noisy social media and collaborative resources.



Arzt, A., Goebl, W., & Widmer, G. (2015). Flexible score following: The Piano Music Companion and beyond. In A. Mayer, V. Chatziioannou, & W. Goebl (Eds.), Proceedings of the 3rd Vienna Talk on Music Acoustics: Institute of Music Acoustics, University of Music and Performing Arts Vienna.

Dingjan, M., Kapel, R., & Sijm, M. (2017). An Automated Language Translation System for Classical Music Titles. BSc thesis [supervisor: C. C. S. Liem]. Delft, The Netherlands: TU Delft.

Eskevich, M., Jones, G.J., Aly, R., Ordelman, R.J., Chen, S., Nadeem, D., Guinaudeau, C., Gravier, G., Sébillot, P., De Nies, T., and Debevere, P. “Multimedia information seeking through search and hyperlinking”, Proc. ICMR, 2013.

İren, D., Liem, C.C.S., Yang, J., and Bozzon, A. “Using social media to reveal social and collective perspectives on music,” Proc. WebSci, 2016.

Liem, C.C.S. “From Water Music to ‘Underwater Music’: Multimedia Soundtrack Retrieval with Social Mass Media Resources,” Proc. TPDL, 2016.

Sutcliffe, R. et al. “The C@merata task at MediaEval 2016: Natural Language Queries Derived from Exam Papers, Articles and Other Sources against Classical Music Scores in MusicXML”, Working Notes Proc. MediaEval, 2016.