Home » Seen and Liked

Recent MIR theses on chords, keys and harmony

6 März 2011 1,386 views No Comment

The past year has been a great one for harmony-related MIR. The chord detection task at MIREX has arguably been more popular than ever, and new ground truth annotation projects have started at McGill et al. Other projects, however, have come to a well-deserved end: a bunch of PhD theses (please let me know if I’ve missed any, or even yours).

  • Christopher Harte: Towards automatic extraction of harmony information from music signals QMUL library link,
  • Matthias Mauch: Automatic chord transcription using computational models of musical context on this website,
  • Laurent Oudre: Template-based chord recognition from audio signals PDF at Telecom Paris Tech,
  • Hélène Papadopoulos: Joint Estimation of Musical Content from an Audio Signal PDF at Archives Ouvertes.

In the worst case, these will be quite a nice reference for those about to finish their own thesis. In the best case, there’ll be something to learn from them. What is definitely interesting is that these four have managed to cover what’s essentially the same topic in very different ways. Hélène and myself have very much relied on Bayesian models and context, while the other two have focused more on the actual shape of chord patterns and evaluation. But don’t trust my judgement, read for yourselves.

In a vain attempt of completeness, here’s a bunch of older theses on harmony in MIR, collected on Elias Pampalk’s MIR PhDs page:

  • Kyogu Lee: A System for Acoustic Chord Transcription and Key Extraction from Audio Using Hidden Markov Models Trained on Synthesized Audio (2008) PDF from Stanford,
  • Katy Noland: Computational Tonality Estimation: Signal Processing and Hidden Markov Models (2009) PDF at QMUL,
  • Jeremy Pickens: Harmonic Modeling for Polyphonic Music Retrieval (2004),
  • Matti Ryynänen: Automatic Transcription of Pitch Content in Music and Selected Applications (2008) PDF.

Comments are closed.