Improving Timbre Similarity : How high’s the sky ?
Many variants that surprisingly do not lead to any substancial improvement are described, suggesting the existence of a “glass ceiling” at R-precision about 65% which cannot probably be overcome by pursuing such variations on the same theme.
Music Similarity Measures: What's the use?
- J. Aucouturier, F. Pachet
- Computer ScienceInternational Society for Music Information…
- 13 October 2002
A timbral similarity measures for comparing music titles is introduced based on a Gaussian model of cepstrum coefficients and it is shown that the measure is able to yield interesting similarity relations, in particular when used in conjunction with other similarity relations.
The Continuator: Musical Interaction With Style
- F. Pachet
- Computer ScienceInternational Conference on Mathematics and…
- 1 September 2003
The Continuator is a system that bridges the gap between two classes of traditionally incompatible musical systems, based on a Markov model of musical styles augmented to account for musical issues such as management of rhythm, beat, harmony, and imprecision.
Representing Musical Genre: A State of the Art
This article discusses the various approaches in representing musical genre, and proposes to classify these approaches in three main categories: manual, prescriptive and emergent approaches.
The bag-of-frames approach to audio pattern recognition: a sufficient model for urban soundscapes but not for polyphonic music.
- J. Aucouturier, Boris Defreville, F. Pachet
- Computer ScienceJournal of the Acoustical Society of America
- 31 July 2007
This paper proposes to explicitly examine the difference between urban soundscapes and polyphonic music with respect to their modeling with the BOF approach, and reveals critical differences in the temporal and statistical structure of the typical frame distribution of each type of signal.
Deep Learning Techniques for Music Generation - A Survey
This paper is a survey and an analysis of different ways of using deep learning (deep artificial neural networks) to generate musical content, based on the analysis of many existing deep-learning based systems for music generation selected from the relevant literature.
A taxonomy of musical genres
This work describes a novel music genre taxonomy based on a few guiding principles, and reports on the process of building this taxonomy.
Hit Song Science Is Not Yet a Science
The experiment shows that some subjective labels may indeed be reasonably well-learned by these techniques, but not popularity, which contradicts recent and sustained claims made in the MIR community and in the media about the existence of “Hit Song Science”.
A scale-free distribution of false positives for a large class of audio similarity measures
FINDING SONGS THAT SOUND THE SAME
A technique that allows users to find music titles that sound similar to songs they like by relying on a modelling of the timbral characteristics of a music by distributions of Cepstrum coefficients and an evaluation of the quality of the extracted similarity measure is proposed.