Data Mining for Alternate Fingerings and Multiphonics of the Modern Flute - Andrew Botros BE Thesis 2001

Winner Siemens Prize for Innovation 2002
Winner Australian Acoustical Society Excellence in Acoustics Award 2003

Andrew M. Botros
Bachelor of Engineering (Computer) Thesis 2001
University of New South Wales

Supervised by Prof. Joe Wolfe, School of Physics UNSW.

Click here for the thesis end product...

The Virtual Boehm Flute

·  Complete thesis 5.4M PDF (ZIP)
·  Introductory summary 124K PDF
·  Selected references
·  Thesis abstract


Selected References

BOTROS, A., SMITH, J. & WOLFE, J. (2006). The Virtual Flute: An advanced fingering guide generated via machine intelligence. Journal of New Music Research, 35(3), 183-196.
Article (1.1M PDF)

CAVANAGH, J. (2005). Find 40,000 new flute fingerings! Pan (Journal of the British Flute Society), 24(2), 14-17 (June 2005).
Article (1.1M PDF)

BOTROS, A., SMITH, J. & WOLFE, J. (2002). The Virtual Boehm Flute — a web service that predicts multiphonics, microtones and alternative fingerings. Acoustics Australia, 30, 61-65.
Article (449K PDF)

BOTROS, A., SMITH, J. & WOLFE, J. (2003). 'The Virtual Flute': acoustic modelling at the service of players and composers. Proc. Stockholm Music Acoustics Conference (SMAC 03), Stockholm. R. Bresin, ed., pp. 247-250.
Article (86K PDF) | Presentation (175K PDF)

BOTROS, A., SMITH, J. & WOLFE, J. (2001). Alternative fingerings and multiphonics: measurements, analyses and databases. Proc. International Symposium on Musical Acoustics, Perugia. D. Bonsi, D. Gonzales, D. Stanzial, eds., pp. 509-512.

BOTROS, A. (2002). The Virtual Flute: a web service for musicians. Australian Music and Psychology Society Seminars, 6 September, University of New South Wales, Sydney.
Abstract | Presentation (1.1M PDF)

WINNER, Siemens Prize for Innovation, 2002. "An award for Australia's most outstanding final year student project in electrical, electronic, communication or computer engineering."
Presentation (118K PDF) | Siemens press release (66K PDF)

"A fingering map for virtual virtuosos," Jeremy Roberts, The Australian, 8 May 2002. Image (218K JPG)

"'Cyberflute' wins national innovation prize," UNSW Media & Communications, 16 May 2002. Article

"Siemens in tune with university students," Matthew Denby, Australian Electronics Engineering, 9 May 2002. Article

WINNER, Australian Acoustical Society Excellence in Acoustics Award, 2003. Jointly sponsored by CSR Bradford Insulation. "This award aims at fostering and rewarding excellence in acoustics, and entries are judged on demonstrated innovation from within any field of acoustics."
Submission (188K PDF) | Presentation (937K PDF) | CSR Bradford Insulation press release


Abstract

      The modern flute can be played in 39,744 acoustically unique configurations (or fingerings). Advanced flute books give hundreds of alternate fingerings and multiphonics (i.e. chords, where the flute sounds two or more notes simultaneously). In practice there are many thousands of unknown and potentially useful fingerings.
      Instruments in the flute family, unlike most wind instruments, are played with the input of the instrument open to the atmosphere. Consequently, they operate at minima in the spectrum of acoustic input impedance. The development of an accurate acoustic input impedance simulation has the potential to predict many of the features of the modern flute’s entire musical response.
      Several researchers over the last few decades have made one-dimensional waveguide models to calculate impedance spectra of woodwind instruments. Past models of the flute have not had the benefit of accurate impedance measurements to allow accurate determination of the required parameters. In the present study, systems of increasing complexity were measured to determine parameters one at a time, and the complete model was tested against accurate impedance measurements made by the UNSW Music Acoustics laboratory. Because of the differences between instrumental geometries and those whose acoustic properties are given by exact theory, several parameters appear that, although not arbitrary, must necessarily be adjusted to fit measured data. Nine such parameters are identified and adjusted, and the resulting simulated impedance spectra agree well with the many measured spectra available. To predict the playability of acoustic impedance minima, an expert system is built using the experimental results of a good flutist. The playability of 957 impedance minima are evaluated and ranked on an arbitrary scale of 0 - 3. The resulting C5.0 decision tree and Cubist linear model establish a relation between the physical parameters of impedance minima and the playable notes of a fingering. The notes and multiphonics of all flute fingerings can thereafter be evaluated to a reasonable degree of accuracy.
      The predicted results of impedance calculations and data analysis are presented as a musically useful web service. Alternate fingerings and multiphonics are stored in a large relational database, searchable in a structured manner. The Virtual Boehm Flute is permanently located at
http://www.phys.unsw.edu.au/music/flute/virtual/