Output details
35 - Music, Drama, Dance and Performing Arts
University of Durham
SuperCollider Music Information Retrieval Library
The SuperCollider Music Information Retrieval Library (SCMIR) is an extension set for the SuperCollider audio programming language for automatic music analysis tasks on recordings. It supports non-realtime (faster than realtime) audio file corpus analysis for computational musicology, as manifest in the growing field of content-based music information retrieval. The software includes feature extraction (audio analysis/machine listening) components, as well as machine learning facilities. SCMIR can be used for batch processing of large databases of audio files as well as single files; typical larger scale projects might run over hundreds of audio files. Compositional applications can also be derived from such analysis, including realtime systems built on principles of machine listening and learning which draw on large databases for their knowledge.
Included here as items in the portfolio are:
(a) The software itself, also available from http://www.dur.ac.uk/nick.collins/code/SCMIR.zip
(b) A paper describing the software:
(2011) SCMIR: A SuperCollider Music Information Retrieval Library. Proceedings of ICMC2011, International Computer Music Conference, Huddersfield
(c) A paper describing a project which used the software:
(2012) Influence in Early Electronic Dance Music: An Audio Content Analysis Investigation. International Society for Music Information Retrieval conference (ISMIR), Porto.
(d) A second paper describing a project which used the software:
(2013) Noise Music Information Retrieval, in Aaron Einbond and Aaron Cassidy (eds.) Noise in Music book, University of Huddersfield Press.
(e) An audio demo and programme notes of the interactive musical artificial intelligence Schubot, based on a corpus of Schubert and Finnissy piano works, as premiered in a concert at ISMIR 2012 in Porto (10 October) and subsequently as part of a recital at the Hochschule für Musik, Karlsruhe, 15 November 2012. Schubot’s processing changes, based on its detection of Schubertian piano music or otherwise.
The library additionally supported elements of the Autocousmatic project mentioned under Collins’ first submission.