ASyMMuS at Lorentz Center Leiden Workshop on Music Similarity

Several researchers from the ASyMMus and DML projects prominently contributed to the high-profile international workshop “Music Similarity: Concepts, Cognition and Computation“.

The workshop gathered experts on music similarity from Computer Science, Musicology, Music Psychology and related scientific fields. In a highly-motivated series of workgroups and talks, our researchers collaborated with other experts in the field in theoretical concepts and computer models of music similarity.

Main areas that were addressed :
* Relationship of similarity and categorisation
* Embedding similarity in context
* Perception and cognition of similarity
* Similarity modelling
* Similarity in music content – music analysis
* Similarity in music expression

Results include a roadmap for interdisciplinary music similarity research as well as future collaborations across scientific fields.

DML and ASyMMuS projects at DMRN+9 workshop

Current progress on the DML and ASyMMuS projects will be presented at the Digital Music Research Network Workshop 2014 (DMRN+9), taking place on Tuesday 16th December at Queen Mary University of London. The list of project-related presentations is as follows:

  • “The ASyMMuS project: An integrated audio-symbolic model of music similarity”, Emmanouil Benetos, Daniel Wolff, Tillman Weyde (City University London), Nicolas Gold, Samer Abdallah (University College London) and Alan Marsden (Lancaster University)
  • “Towards analysing big music data – Progress on the DML research project”, Tillman Weyde, Stephen Cottrell, Jason Dykes, Emmanouil Benetos, Daniel Wolff, Dan Tidhar, Alexander Kachkaev (City University London), Mark Plumbley, Simon Dixon, Mathieu Barthet, Steven Hargreaves (Queen Mary University of London), Nicolas Gold, Samer Abdallah (University College London), Aquiles Alancr-Brayner, Mahendra Mahey and Adam Tovell (The British Library)

DML at UCL Digifest

The first UCL Digifest was held last month, from 10th–14th November. Digifest is “a celebration of all things digital at UCL”, which for the UCL Music Systems Engineering Resarch Team (a.k.a MUSERT) , a.k.a Nicolas Gold, Samer Abdallah and Christodoulos Aspromalis) meant a chance to show off some of our recent activities as well as to call together the first meeting and performance of the UCL Laptop Orchestra, or UCLOrk.

The UCLOrk meeting (on Wednesday 12th) consisted of a tutorial session on computer music and building digital instruments using PureData, followed by a performance. The Computer Science technical support team worked wonders by building (in a very short space of time) 12 hemispherical speakers, so that each member of the orchestra could have their own speaker next to them—hemisphercal speakers are often used for laptop orchestras as they diffuse the sound better than ordinary speakers and mimic the effect of having many instruments distributed physically around the performance space. At the end of the session, orchestra members to their cues from an animated visual score created by Christodoulos—you can find a video here (complete with the occasional Mac volume changing sound…).

At the showcase session on Friday 14th, we demonstrated various music-related applications, including Christodoulos’s affective generative music system for computer games, the MiCLUES app (MiCLUES is a Share Academy/Arts Council England-funded project in collaboration with the Royal College of Music (RCM)) to guide museum visitors to the Museum of Instruments at the RCM., a device to help keep to the recommended 4 minute shower time limit, and a prototype of the DML information management system. The DML prototype allowed users to browse an RDF graph containing information about a local collection of audio files and symbolic scores, and then use this as a jumping off point for going out into the Semantic Web to pull in more information (via Linked Open Data and SPARQL endpoints), for example, from MusicBrainz (and LinkedBrainz) or DBpedia. Items in the symbolic music library could be accessed as machine readable scores (in several formats, such as MIDI, Humdrum/Kern, MusicXML, Lilypond), traditionally engraved scores (using Lilypond), or audio rendered from MIDI using Fluidsynth. The prototype also showed how a computational engine, (in this case using Matlab) can be integrated into the system, so that large scale musicological research can be conducted and the results managed.

Finally, running throughout the week was an application to collect visitor feedback, in the form of descriptive or mood-related words, and generate a music playlist reflecting the feedback. The playlist was compiled by using The Echonest web service to search for songs matching a subset of descriptive words, which were then added (after filtering out anything likely to be too offensive!) to a Spotify playlist using the Spotify API. If you have a spotify account, you can see the last state of the playlist under user name ‘ucldigifest’, playlist name ‘digifest’. It may change without warning at any time!

DML project at CIM 2014 conference

Current progress on the DML project will be presented at the 9th Conference on Interdisciplinary Musicology (CIM 2014). CIM will take place on 4-6 December in Berlin, Germany. Project-related papers are listed below (click titles to download abstracts):

DML project at Musical Timbre Workshop

cropped-T08-violin_cepstrum_spenvelopeRecent work on instrumentation recognition that was carried out as part of the DML project was presented at the Workshop on Musical Timbre, that took place on 14th November at Télécom ParisTech, in Paris, France. The talk was entitled “Instrument transcription & instrumentation recognition”.

For more information on our work in instrumentation recognition, please see our DLfM paper:

DML project at the Society for Ethnomusicology Annual Meeting

Current progress on the DML project will be presented by Stephen Cottrell at the 59th annual meeting of the Society for Ethnomusicology, taking place on 13-16 November in Pittsburgh, Pennsylvania. The abstract for the talk can be found below:

Ethnomusicology, Music Information Retrieval and Big Data

The application of Music Information Retrieval (MIR) techniques to large recorded corpuses of music beyond the Western traditions, taken at face value, appears to run contrary to many of the principles which ethnomusicologists have long held dear. Culturally decontextualised, and with all the problems that attach to specific recordings serving as single instantiations of otherwise diverse musical traditions, the large-scale computerised analysis of such recordings risks appearing to return the discipline to its comparative musicology roots. Nevertheless, some studies have been undertaken over recent years as part of what is sometimes termed ‘computational ethnomusicology’. What is gained is the possibility not only of the kinds of technologically-enhanced analytical exactitude which computers can provide, but also of large-scale comparative analyses both within and across cultures. These offer, for example, the prospect of revisiting on a more scientific basis earlier debates about human universals in music making, in addition to the more usual traits of pitch or melody extraction, semantic categorisation and similar. Thus far, however, MIR studies have generally concentrated on individual recordings or small collections, which do not comfortably facilitate large-scale comparisons. This paper will review some of the latest work being undertaken in the MIR field in relation to global music traditions, and consider the problems and possibilities such approaches present. It will also report on the latest results from an ongoing government-funded UK research project – the Digital Music Lab – which seeks to apply MIR techniques to musical Big Data, specifically focused on music beyond the Western traditions.

Early Music paper on large-scale temperament estimation

As part of the DML project, a paper on large-scale analysis of harpsichord temperament has been authored by Dan Tidhar, Simon Dixon, Emmanouil Benetos, and Tillman Weyde. The paper is entitled “The Temperament Police“, and has been published at the special issue of Early Music (Oxford University Press) on “Early Music and Modern Technology“.

You can view the open access paper as either PDF or HTML (thanks to the City University London Open Access Block Grant for supporting this open access publication).

UPDATE: There is also a supplementary webpage accompanying the article, which is located at

DML project at BL Labs Symposium 2014

BL-labs-logoCurrent progress and future directions of the DML project will be presented by Daniel Wolff and Adam Tovell at the British Library Labs Symposium 2014, taking place on Monday 3rd November 2014, at the British Library Conference Centre. More information on the DML presentation can be found below:

Digital Music Lab – Analysing Big Music Data
Daniel Wolff, Research Fellow for the Digital Music Lab project at the Music Informatics Research Group, City University London and Adam Tovell, Curator, Digital Music, British Library
The Digital Music Lab is developing research methods and software infrastructure for exploring and analysing large-scale music collections, and to provide researchers and users with datasets and computational tools to analyse music audio, scores and metadata. The team will present some of their initial findings including some startling visualisations.

DML project at ISMIR 2014

Current work on the DML project was presented at the 15th International Society for Music Information Retrieval Conference (ISMIR 2014, Taipei, Taiwan):

DML project at AES ‘Cutting Edge Research’ event

The DML project was presented at the event sponsored by the Audio Engineering Society, entitled “Cutting Edge Research – from City University and King’s College London”, which took place at City University on 14th October.

The event showcased cutting edge research from City University’s Music Informatics Research group and King’s College London’s Centre for Telecommunications Research. As part of the event, Tillman Weyde gave a talk on the group’s activities (including the DML project), and Dr Dan Tidhar presented the poster entitled “Big Data for Musicology and Music Retrieval”.