playlist generation. Motivation. Sebastian Streich

1
hgkjhgjhg
Estimating the Perceived Complexity of
Music Sebastian Streich Universitat Pompeu
Fabra Prof. Xavier Serra

• hgljhglhjg

2
Contents of this presentation

• An Introduction to Music Complexity
• Motivation
• What can I do?
• Existing Models
• My Own Approaches
• Future work

3
Music Complexity

• Teaser Spot the complexity!

4
Music Complexity

• Unusual harmonies and timbres, irregular tempi
  and rhythms,
• unexpected tone sequences and variations in
  volume
• raise the level of perceived music complexity.
  Finnäs 1989
• expectation and surprise, variation, and
  information rate have influence on perceived
  complexity
• partly subjective (experience, cultural
  background, )
• several different, independent aspects of musical
  complexity
• multi-dimensional representation
• link to high-level and low-level musical features

5
Music Complexity

• Six dimensions

6
Motivation

• What is it good for?
• searching in large song collections
• e.g. songs with simple melodies/interesting
  rhythm/
• enhancing clustering methods (in combination with
  other features)
• visualise song collections
• complexity related to
• preference Berlyne 1971
• user complexity profile/
• music recommendation/
• playlist generation

7
What can I do?
typical music from western digital collections
sound file (song)
8
Complexity Models

• General
• Kolmogorov complexity (information theory)
• length of shortest program complexity
• perceptual relevance?
• (e.g. random sequence max. complexity)
• theoretical measure only
• Complexity of short musical excerpts Scheirer
  2000
• uses five psychoacoustic features extracted from
  audio
• found to be significant in predicting perceived
  complexity of short musical excerpts
• no distinction between different aspects of
  complexity

9
Complexity Models

• Rhythm
• PS-Measure Povel/Shmulevich 2000
• clock induction strenght and segment coding
  complexity
• found to be significant in predicting the
  perceived rhythmic complexity of short patterns
• needs symbolic representation
• Information-based complexity Pressing 1999
• processing cost function for production of
  rhythmic pattern
• no evaluation published
• needs symbolic representation

10
Complexity Models

• Melody
• Expectancy-based complexity Eerola, North 2000
• predictability of the melody (based on tonality,
  intervallic and rhythmic principles)
• found to be significant in predicting the
  perceived melodic complexity of musical excerpts
• needs symbolic representation
• Harmony, Timbre, Structure, Acoustic ?

11
My Own Approaches

• Rhythm
• danceability measure based on loudness
  fluctuation
• combination of beat/onset detector with
  PS-Measure
• Acoustic
• split into spatial and dynamic part
• Timbre
• unsupervised learning of timbre models
• texture description compression

12
Future Work

• Evaluation methods (playlist ordering, song
  rating, ?)
• connect audio MIR methods with existing models
• develope more precise models
• address subjectivity of complexity?

13
References

• L. Finnäs, How can musical preference be
  modified? A research review, Bulletin of the
  Council for Research in Music Education no. 102,
  pp. 158 (1989).
• D. E. Berlyne, Aesthetics and psychobiology,
  Appleton-Century-Crofts (1971).
• E. D. Scheirer, R. B. Watson, B. L. Vercoe, On
  the perceived complexity of short musical
  segments, Proc. of International Conference on
  Music Perception and Cognition, CD-ROM (2000).
• I. Shmulevich, D. J. Povel, Measures of temporal
  pattern complexity, Journal of New Music
  Research vol. 29, no. 1 (2000).
• J. Pressing, Cognitive complexity and the
  structure of musical patterns, Proc. of the 4th
  Conference of the Australasian Cognitive Science
  Society (1999).
• T. Eerola, A. C. North, Expectancy-Based Model
  of Melodic Complexity, Proc. of the 6th
  International Conference on Music Perception and
  Cognition, CD-ROM (2000).