Generative Structures in Improvisation: Computational Segmentation of Keyboard Performances

Dean, Roger T.; Bailes, Freya; Drummond, Jon

doi:10.1080/09298215.2013.859710

Cited by 12 publications

(13 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Musical feature change is a common cue for segmentation: listeners indicate segment boundaries if they easily perceive that there is a contrast, such as a stark change in dynamics or instrumentation. Multiple strategies are exploited by composers (Deliège, 2001), improvisers (Dean, Bailes, & Drummond, 2014) and performers (Poli, Rodà, & Vidolin, 1998) to induce perception of musical changes, and communicate musical structure to the listener. This paper focuses, however, on musical listeners only, and on a particular conception of segmentation.…”

Section: Introductionmentioning

confidence: 99%

Interaction features for prediction of perceptual segmentation: Effects of musicianship and experimental task

Hartmann

Lartillot

Toiviainen

2016

Journal of New Music Research

View full text Add to dashboard Cite

As music unfolds in time, structure is recognized and understood by listeners, regardless of their level of musical expertise. A number of studies have found spectral and tonal changes to quite successfully model boundaries between structural sections. However, the effects of musical expertise and experimental task on computational modelling of structure are not yet well understood. These issues need to be addressed to better understand how listeners perceive the structure of music and to improve automatic segmentation algorithms. In this study, computational prediction of segmentation by listeners was investigated for six musical stimuli via a real-time task and an annotation (non real-time) task. The proposed approach involved computation of novelty curve interaction features and a prediction model of perceptual segmentation boundary density. We found that, compared to non-musicians', musicians' segmentation yielded lower prediction rates, and involved more features for prediction, particularly more interaction features; also nonmusicians required a larger time shift for optimal

show abstract

Section: Introductionmentioning

confidence: 99%

Interaction features for prediction of perceptual segmentation: Effects of musicianship and experimental task

Hartmann

Lartillot

Toiviainen

2016

Journal of New Music Research

View full text Add to dashboard Cite

show abstract

“…Bruderer (2008), Wiering, de Nooijer, Volk, and Tabachneck-Schijf (2009), and Pearce et al (2010) have compared the performance of some segmentation systems. Other work on segmentation includes a neural study on finding working memory triggers (Burunat, Alluri, Toiviainen, Numminen, & Brattico, 2014) and a performance study on improvisational structure (Dean, Bailes, & Drummond, 2014). Outside our scope, work on musical closure has explored the role of musicianship and experience on boundary perception of classical music (Peebles, 2011;Sears, Caplin, & McAdams, 2014).…”

mentioning

confidence: 99%

Multi-scale Modelling of Segmentation

2016

View full text Add to dashboard Cite

While listening to music, people often unwittingly break down musical pieces into constituent chunks such as verses and choruses. Music segmentation studies have suggested that some consensus regarding boundary perception exists, despite individual differences. However, neither the effects of experimental task (i.e., real-time vs. annotated segmentation), nor of musicianship on boundary perception are clear. Our study assesses musicianship effects and differences between segmentation tasks. We conducted a real-time experiment to collect segmentations by musicians and nonmusicians from nine musical pieces. In a second experiment on non-real-time segmentation, musicians indicated boundaries and their strength for six examples. Kernel density estimation was used to develop multi-scale segmentation models. Contrary to previous research, no relationship was found between boundary strength and boundary indication density, although this might be contingent on stimuli and other factors. In line with other studies, no musicianship effects were found: our results showed high agreement between groups and similar inter-subject correlations. Also consistent with previous work, time scales between one and two seconds were optimal for combining boundary indications. In addition, we found effects of task on number of indications, and a time lag between tasks dependent on beat length. Also, the optimal time scale for combining responses increased when the pulse clarity or event density decreased. Implications for future segmentation studies are raised concerning the selection of time scales for modelling boundary density, and time alignment between models.

show abstract

“…We plan to introduce both features into our system later. (1)(2)(3)(4)(5)(6)(7)(8)(9)(10), together with the corresponding single velocity (v), duration (d) and inter-onset interval (ioi) to the next event. When an input event is a 'note', only p1 has a positive number, and p1-10 are each -1.…”

Section: Specific Purposes and The Musical Representationmentioning

confidence: 99%

“…An experienced free improviser in any artistic form is usually capable of responding to highly diverse, often unanticipated inputs in ways that are potentially equally diverse and sometimes unfamiliar even to the improviser [4,6]. Experimental work has revealed something of the decision making involved [7][8][9] (reviewed [10,11]). By the same token, a free improviser may choose for example to adopt a tonal or metrical posture, or engage in blues-oriented phrases, at any time.…”

Section: Introductionmentioning

confidence: 99%

Towards a Deep Improviser: a prototype deep learning post-tonal free music generator

Dean

Forth

2018

Neural Comput & Applic

Self Cite

View full text Add to dashboard Cite

Two modest-sized symbolic corpora of post-tonal and post-metrical keyboard music have been constructed, one algorithmic, the other improvised. Deep learning models of each have been trained. The purpose was to obtain models with sufficient generalisation capacity that in response to separate fresh input seed material, they can generate outputs that are statistically distinctive, neither random nor recreative of the learned corpora or the seed material. This objective has been achieved, as judged by k-sample Anderson-Darling and Cramer tests. Music has been generated using the approach, and preliminary informal judgements place it roughly on a par with an example of composed music in a related form. Future work will aim to enhance the model such that it deserves to be fully evaluated in relation to expression, meaning and utility in real-time performance.

show abstract

Generative Structures in Improvisation: Computational Segmentation of Keyboard Performances

Cited by 12 publications

References 12 publications

Interaction features for prediction of perceptual segmentation: Effects of musicianship and experimental task

Interaction features for prediction of perceptual segmentation: Effects of musicianship and experimental task

Multi-scale Modelling of Segmentation

Towards a Deep Improviser: a prototype deep learning post-tonal free music generator

Contact Info

Product

Resources

About