Analysis of musical structure in audio and MIDI signals using information rate

Shlomo Dubnov

Research output: Contribution to conferencePaperpeer-review


Information rate (IR) was recently introduced as a novel measure of musical structure that quantifies prediction properties of musical signals. Formulated in information theoretic terms, it quantifies the rate of change of information in time series, and is related to capability of a system to perform information analysis of the data, such as actions involved in music listening. In this paper we extend IR measure to the case of MIDI sequences and perform a comparative analysis of audio signals (acoustic recordings) and symbolic representations (note, i.e. MIDI) of same musical piece. We show that in both cases IR detects similar types of structure, suggesting that IR might capture some inherent properties of musical signals.

Original languageEnglish
Number of pages6
StatePublished - 1 Jan 2006
Externally publishedYes
EventInternational Computer Music Conference, ICMC 2006 - New Orleans, United States
Duration: 6 Nov 200611 Nov 2006


ConferenceInternational Computer Music Conference, ICMC 2006
Country/TerritoryUnited States
CityNew Orleans


  • Complexity
  • Entropy
  • Information rate
  • Musical structure
  • Prediction

ASJC Scopus subject areas

  • Media Technology
  • Computer Science Applications
  • Music


Dive into the research topics of 'Analysis of musical structure in audio and MIDI signals using information rate'. Together they form a unique fingerprint.

Cite this