Showing papers by "J. Stephen Downie published in 2012"

Second fiddle is important too: pitch tracking individual voices in polyphonic music

Abstract: Recently, there has been much interest in automatic pitch estimation and note tracking of polyphonic music. To date, however, most techniques produce a representation where pitch estimates are not associated with any particular instrument or voice. Therefore, the actual tracks for each instrument are not readily accessible. Access to individual tracks is needed for more complete music transcription and additionally will provide a window to the analysis of higher constructs such as counterpoint and instrument theme imitation during a composition. In this paper, we present a method for tracking the pitches (F0s) of individual instruments in polyphonic music. The system uses a pre-learned dictionary of spectral basis vectors for each note for a variety of musical instruments. The method then formulates the tracking of pitches of individual voices in a probabilistic manner by attempting to explain the input spectrum as the most likely combination of musical instruments and notes drawn from the dictionary. The method has been evaluated on a subset of the MIREX multiple-F0 estimation test dataset, showing promising results.

How significant is statistically significant? The case of audio music similarity and retrieval

Abstract: The principal goal of the annual Music Information Retrieval Evaluation eXchange (MIREX) experiments is to determine which systems perform well and which systems perform poorly on a range of MIR tasks. However, there has been no systematic analysis regarding how well these evaluation results translate into real-world user satisfaction. For most researchers, reaching statistical significance in the evaluation results is usually the most important goal, but in this paper we show that indicators of statistical significance (i.e., small p-value) are eventually of secondary importance. Researchers who want to predict the realworld implications of formal evaluations should properly report upon practical significance (i.e., large effect-size). Using data from the 18 systems submitted to the MIREX 2011 Audio Music Similarity and Retrieval task, we ran an experiment with 100 real-world users that allows us to explicitly map system performance onto user satisfaction. Based upon 2,200 judgments, the results show that absolute system performance needs to be quite large for users to be satisfied, and differences between systems have to be very large for users to actually prefer the supposedly better system. The results also suggest a practical upper bound of 80% on user satisfaction with the current definition of the task. Reflecting upon these findings, we make some recommendations for future evaluation experiments and the reporting and interpretation of results in peer-reviewing.

Music interaction research in HCI: let's get the band back together

Abstract: The ubiquity of music consumption is overarching. Statistics for digital music sales, streaming video videos, computer games, and illegal sharing all speak of a huge interest. At the same, an incredible amount of data about every day interactions (sales and use) with music is accumulating through new cloud services. However, there is an amazing lack of public knowledge about everyday music interaction. This panel discusses the state of music interaction as a part of digital media research. We consider why music interaction research has become so marginal in HCI and discuss how to revive it. Our two discussion themes are: orientation towards design vs. research in music related R&D, and the question if and how private, big data on music interactions could enlighten our understanding of ubiquitous media culture.

The impact of MIREX on scholarly research (2005-2010)

Abstract: This paper explores the impact of the MIREX (Music Information Retrieval Evaluation eXchange) evaluation initiative on scholarly research. Impact is assessed through a bibliometric evaluation of both the MIREX extended abstracts and the papers citing the MIREX results, the trial framework and methodology, or MIREX datasets. Impact is examined through number of publications and citation analysis. We further explore the primary publication venues for MIREX results, the geographic distribution of both MIREX contributors and researchers citing MIREX results, and the spread of MIREX-based research beyond the MIREX contributor teams. This analysis indicates that research in this area is highly collaborative, has achieved an international dissemination, and has grown to have a significant profile in the research literature.

Reuse, remix, repeat: the workflows of mir

Abstract: Many solutions for the reuse and remixing of MIR methods and the tools implementing them have been introduced over recent years. Proposals for achieving the necessary interoperability have ranged from shared software libraries and interfaces, through common frameworks and portals, to standardised file formats and metadata. Each proposal shares the desire to reuse and combine repurposable components into assemblies (or “workflows”) that can be used in novel and possibly more ambitious ways. Reuse and remixing also have great implications for the process of MIR research. The encapsulation of any algorithm and its operation ‐ including inputs, parameters, and outputs ‐ is fundamental to the repeatability and reproducibility of any experiment. This is desirable both for the open and reliable evaluation of algorithms (e.g. in MIREX) and for the advancement of MIR by building more effectively upon prior research. At present there is no clear best practice widely adopted throughout the community. Should this be considered a failure? Are there limits to interoperability unique to MIR, and how might they be overcome? In this paper we assess contemporary MIR solutions to these issues, aligning them with the emerging notion of Research Objects for reproducible research in other domains, and propose their adoption as a route to reuse in MIR.