Algorithms for Computer Music



When: Summer semester 2018

Contact:

Mickael Zehren, Prof. Paolo Bientinesi
Send an email

Overview

This seminar is well suited for motivated students who are passionate about both music and computers. Ideally, you want to work on a MS thesis in computer music.

All the topics are concerned with the algorithmic/automatic solution of a music-related problem.
E.g.: identification of a musical composer, generation of music, simulation of an acoustic environment, ...

Every student gives a presentation and writes a report on the assigned topic. The presentation must contain music.


Prerequisites

  • Reasonable knowledge of the fundamentals of music
  • Good command of algorithms & complexity
  • Familiarity with multiple programming languages
  • Strong mathematical background


Seminar rules

  • Attendance is mandatory
  • Grade:
    • 70% presentation, Q&A
    • 30% report


Topics

  • MIR: Music Information Retrieval
  • KD: Key Detection
  • CR: Chord Recognition
  • ME: Melody extraction
  • MT: Automatic Music Transcription
  • OD: Onset Detection
  • BT: Beat tracking
  • MSA: Music Structure Analysis
  • SM: Automatic Sound Mixing
  • MC: Music Composition

Assignments

# Name Topic Publication Implementation Suggested exercise
1 Marc Burian MIR PDF GitHub Try the main algorithms
2 Paramananthan Prethep MIR PDF GitHub Try the main algorithms
3 Wei Lu KD PDF GitHub Try the trained model and replicate the results
4 Tobias Winkler CR PDF GitHub Try the trained model and replicate the results
5 Lia Schulze Dephoff MT PDF Website Analyse the models and the tasks
6 Simon Convent MT PDF GitHub Try the trained model and replicate the results
7 Jan-Frederik Konopka OD PDF Not available Implement the method
8 Simon Veittes BT PDF Not available Implement the method
9 Tobias Möhring MSA PDF GitHub Replicate the results of figure 1 and figure 2

Task

The publication assigned to you is what you will present in class. You may want to also look at other resources to learn more details about the main topic and/or about methods used in the initial paper.

You want to present the problem, approach, results. Have a critical eye: explain why the approach is interesting (what does it bring, what is the motivation) and criticise it (what do you think is particularly well done or not, how you would improve it).

Your presentation will last 20-25" with 5" of Q&A. It should contain music, examples, tests, schemes or anything which can be used to illustrate your topic and ease its comprehension.

Finally, you will have to put into practice the topic presented in the paper. This part is up to you and should be presented briefly in the presentation and in the report. Use this part to enhance your comprehension of the topic or to create examples for your presentation. You can find suggested exercises you could work on in the table. You may want to use the implementation when available for this part. Your plan of action, due to Wednesday 25/04, has to define the exercise you want to do as well as how you would do it.

N.B.: The students who have only their initials in the name column have to send me an email with their full name as well as their matriculation number and study program.

Lectures

  • Start: Wednesday, April 11, 5.15pm
  • Where: Rogowski 115, AICES seminar room, Schinkelstrasse 2

Schedule

  • Wednesday, 11.04 Topic assignment
  • Wednesday, 25.04 Plan of action

    1-page document, specifies the actions you will take for the presentation and the implementation/put into practice.
  • Wednesday, 02.05 Dropout Limit date
  • Wednesday, 23.05 Slides - draft
  • Tuesday, 03.07 Slides - final version
  • Wednesday, 04.07 Presentations
    • Marc Burian
    • Prethep Paramananthan
    • Wei Lu
  • Wednesday, 11.07 Presentations
    • Tobias Winkler
    • Lia Schulze Dephoff
    • Simon Convent
  • Wednesday, 18.07 Presentations
    • Jan-Frederik Konopka
    • Simon Veittes
    • Tobias Möhring
  • Wednesday, 25.07 Report - deadline

    Template (page limit inside)


Material

  • "HOWTO: Presentations", by P. Bientinesi [ PDF]