Research

Lorin Grubb
Graduate Fellow
School of Computer Science
Carnegie Mellon University


Home ] Contact Information ] [ Research ] Computer Music ] Useful Links ]


Recent Research Projects

  • STOCHASTIC SCORE FOLLOWING FOR VOCAL ACCOMPANIMENT
    Automated accompaniment systems take a musical score as input, listen to live performers, and

    play the remaining parts of the score to provide synchronous, real-time accompaniment. These
    systems must track the live performers by determining their score position and tempo in real-time.

    Tracking a vocal performer has proved to be an extremely challenging problem. This work focuses
    on defining and implementing a statistical model for tracking vocal performers. The model is based
    upon observed properties of vocal performance. It uses a variety of information including estimated
    tempo, observed fundamental pitch, and observed spectral features that correlate with speech.
    The model has been fit to measurements taken from actual performances given by vocal students.
    Initial application of this model has demonstrated accompaniment ability that is qualitatively superior
    to previous vocal accompaniment systems. Current efforts focus on quantitatively evaluating how
    well the fitted model can track singers.
      Stochastic Vocal Tracking (postscript)

  • COMPUTER PERFORMANCE WITHIN A MUSICAL ENSEMBLE
    Prior to this work, researchers had implemented several computer systems for accompanying

    a solo performer. The system developed for this project performed as part of a live ensemble, listening
    to and accompanying multiple performers. Using previously developed tracking technology,

    the system independently estimated score position and tempo of each performer in the ensemble.
    Individual estimates were then combined to estimate the ensemble score position and tempo. These
    estimates were used to control the computer performance. The completed system performed works by
    Handel and Mozart with ensembles having as many as four live musicians.
    Ensemble Performance (postscript)

  • Related Research: See the Computer Music Project at CMU.


Selected Publications

Grubb, L. and Dannenberg, R.B.  1997.
"A Stochastic Method of Tracking a Vocal Performer."
In Proceedings of the1997 International Computer Music Conference (ICMC-97), 301-308.
San Francisco: International Computer Music Association.

Grubb, L. and Dannenberg, R.B.  1995.  (Video)
"Ensemble Accompaniment."
In Computer Music Video Review, 2(1).
San Francisco: International Computer Music Association (ICMA).

Grubb, L. and Dannenberg, R.B.  1994.
"Automating Ensemble Performance."
In Proceedings of the 1994 International Computer Music Conference (ICMC-94), 63-69.
San Francisco: International Computer Music Association (ICMA).

Grubb, L. and Dannenberg, R.B.  1994.
"Automated Accompaniment of Musical Ensembles."
In Proceedings of the Twelfth National Conference on Artificial Intelligence (AAAI-94), 94-99.
Cambridge, MA: MIT Press.

Grubb, L. and Dannenberg, R.B.  1994.
"Computer Performance in an Ensemble."
In Third International Conference for Music Perception and Cognition Proceedings, 57-60.
Liege, Belgium: European Society for the Cognitive Sciences of Music.

Grubb, L.  1992.
"Use of Computational Models in a Rule-Based Expert Decision System."
In Proceedings of the 1992 IAKE Symposium on Knowledge Engineering, 597-602.
Gaithersburg, MD: International Association of Knowledge Engineers (IAKE).

Grubb, L.  1992.
"Domain Independent Natural Language Processing and Automated Reasoning."
In Proceedings of the 1992 IAKE Symposium on Knowledge Engineering, pp. 754-765.
Gaithersburg, MD: International Association of Knowledge Engineers (IAKE).


email: lgrubb@cs.cmu.edu                                                                           Last updated March 13, 1998