python for manufacturing musical instruments
play

Python for manufacturing musical instruments Olivier CAYROL - June - PowerPoint PPT Presentation

Python for manufacturing musical instruments Olivier CAYROL - June 15th, 2016 2 Prolegomena A few words about me: co-founder and deputy CEO of Logilab cuddling computers for 30 years data modelling, software design


  1. Python for manufacturing musical instruments Olivier CAYROL - June 15th, 2016

  2. 2 Prolegomena • A few words about me: • co-founder and deputy CEO of Logilab • cuddling computers for 30 years • data modelling, software design • A few words about my company: • created in 2000, 20 engineers today • Python since the beginning, agile development • active free software supporter • development of strategic semantic Web applications: • data.bnf.fr ( national library ), francearchives.fr ( national archives )

  3. 3 Context • Making and repair of musical instruments : • 2,400 companies in France, 2/3 with only 1 employee • employees: 11,000 people in France • turnover: 800 millions Euros / year • exported production rate: 80% • Assets: • traditional craftsmanship • world-famous quality • Challenge: • aggressive competition from foreign countries

  4. 4 Context illustration Restoration of baritone saxo from 1901, Gaëtan Schneider

  5. 5 ITEMM laboratory • European Technological Institute for Musical Professions • Based in Le Mans, France • Designing digital tools dedicated to the manufacturing of musical instruments: • characterization of instruments • analysis of the sounds the instruments produce • development of digital models to predict the sounds of the instruments • Driving the digital revolution of the French sector of instruments craftsmanship

  6. 6 Needs of the instrument makers • When designing an instrument, makers are interested in: • tuning • timbre • ease of playing • Traditionally, makers: • build multiple prototypes • in order to choose the proper instrument dimensions

  7. 7 PAFI platform • Web application developped by ITEMM and Logilab • source code to be soon published as free software • Support for woodwind and brass wind instruments: • trumpets, horns, trombones, saxophones, clarinets, oboes, etc. • Dedicated digital tools: • instrument models, fingering descriptions • computation of acoustic input impedance for a given fingering • tuning diagram for each fingering • capture of acoustic input impedance from a real instrument • Ability to share and clone instrument models

  8. 8 PAFI platform overview

  9. 9 Use case 1 - introduction • Project made by Baptiste Le Guillou • student at the ITEMM • in the context of his degree in Arts and Crafts ( Brevet des Métiers d'Art ) • project duration: 2 years • Turning a valves trumpet into a valves and slide trumpet • the added slide must modify the note up to one tone • the added slide allows playing quarter tones or glissando effects

  10. 10 Use case 1 - work to be done • Add tubings in the tuning slide to allow the desired effect • Remove a part of the tubing between the bell and the valves • in order to have a longer space for the tuning slide

  11. 11 Use case 1 - study • Questions to be answered: • what length of tubing should be added to the tuning slide? • what are the effects of the modifications on the instrument? • Use of the digital simulation instead of numerous trial / error cycles: • cheaper, faster • ability to explore more possibilities

  12. 12 Use case 1 - geometry • Measure of the actual instrument: • tubings length and diameter • valves position • Description of the instrument in the platform: • series of tubings, cones, valves, returns, holes, etc. • different fingerings: • which holes are closed, half-closed, opened • which valve pistons are pushed • what note is expected

  13. 13 Use case 1 - computation • Computation of the acoustic impedance: • Fourier transform of the pressure divided by the volume flow • shows the resonance frequencies of the instrument • and thus the notes that can be played • Computation of the tuning diagram: • difference between the expected note and the actual note (computed above) • the musician must adjust his playing to correct the note: • small differences mean ease of playing

  14. 14 Use case 1 - results

  15. 15 Use case 1 - checking • Measure of the actual acoustic impedance of the instrument • with dedicated sensors (loudspeakers and microphone) • connection through the Web browser thanks to the Web audio API • Comparison of the two impedance graphs • especially the resonance frequencies

  16. 16 Use case 1 - measures

  17. 17 Use case 1 - conclusion • Different slide lengths have been simulated • The best option has been implemented on the instrument • The result perfectly meets the initial requirements

  18. 18 Use case 2 - introduction • Project made by Gaëtan Schneider • student at the ITEMM • in the context of his degree in Arts and Crafts ( Brevet des Métiers d'Art ) • project duration: 2 years • Restoring a Couesnon baritone saxo from 1901 • numerous keys are twisted • the neck is smashed and splitted • it can't be restored

  19. 19 Use case 2 - initial state

  20. 20 Use case 2 - study • A new neck must be built • Question to be answered: • what must be the length and the shape of the neck? • Study: • geometry description • computation of tuning diagram • digital try of different necks

  21. 21 Use case 2 - model

  22. 22 Use case 2 - results

  23. 23 Under the hood • Web application coded in Python and javascript • Computations based on numpy and scipy libraries • refactored from Matlab prototypes • Simple model and simple computations • easy to describe • fast to compute • sufficient accuracy for the expected physical values • the played note corresponds to the first order in physics

  24. 24 Feedbacks • Craftsmen: • easy test of instrument designs with a digital tool • ability to explore and dig in several options • no installation (Web application) • Researchers: • easy-to-understand code • Python compactness, high-level operations in its numeric libraries • structuring in modules • security from the numerous automatic tests • ability to enhance the algorithms without regression

  25. 25 We are hiring! • Visit http://www.logilab.fr/emplois • Web developer (javascript + Python) • Developer for data analysis and semantic Web (Python) • Thank you for your attention • Email: olivier.cayrol@logilab.fr • Twitter: @OCayrol

Recommend


More recommend